fix: abstain on translucent gradients over images, drop phantom color-mix stops

Two follow-up review findings on the merge with main.

A gradient leading a url() layer was treated as a resolvable surface even
when its stops are translucent, so the glow and AI-palette hunts averaged
wash stops (a 20% black wash reads as pure black) while the real surface
blends with image pixels the engine cannot read. resolveBackgroundInfo now
marks gradient-over-image unresolved unless every readable stop of the
leading gradient is opaque, in which case the gradient provably covers the
image and remains the scorable surface.

parseGradientColorsModern predated this branch's parseGradientColors
rewrite: its second regex pass re-extracted color tokens nested inside
color-mix() stops that the shared parser already captures whole via
balanced-paren tokens, appending ingredient colors that are never painted.
The worst-case stop ratio then invented low-contrast findings against a
color nobody sees. The helper is removed; all callers use the shared
parser, which covers the modern syntaxes it existed for.

Fixture coverage pins both: the translucent-wash-over-image glow abstains
in both engines, an opaque gradient over an image still flags in the
browser, and the color-mix wash case stays clean in the static engine.
Each new assertion was verified to fail against the previous engine.

Addresses Greptile and Cursor Bugbot review findings on PR #541.

AI-assisted-by: Claude Code

Co-Authored-By: Claude <noreply@anthropic.com>

.agents/skills/impeccable/SKILL.md

@@ -9,11 +9,11 @@ This skill gives you the tools and permission to create design that earns to be
 Core principles:
 - Go all out. No hedging, no shortcuts. The deliverable must be complete (except assets the user must provide).
 - Dream big and bold. Distinct, beautiful, outstanding and highly inspiring work.
-- Verify in bounded passes, not a loop, and the ceiling covers the whole cycle: screenshots, defect scans, micro-edits, and rebuilds alike. Build fully, inspect once with a batched round (desktop and mobile together), fix everything it shows in one batch, confirm with at most one more round, and stop polishing. Open-ended self-QA burns the user's money doing worse what the finish handoffs do better.
+- Verify in bounded passes, not a loop, and the ceiling covers the whole cycle: screenshots, defect scans, micro-edits, and rebuilds alike. Build fully, inspect once with a batched round (desktop and mobile together on the web; the shipped device classes on a native platform), fix everything it shows in one batch, confirm with at most one more round, and stop polishing. Open-ended self-QA burns the user's money doing worse what the finish handoffs do better.
 
 ## Setup
 
-1. Run `node .agents/skills/impeccable/scripts/context.mjs` once per session (if the runtime shows this skill's loaded base directory, run `node <skill-base-dir>/scripts/context.mjs`; keep cwd at the user's project). Pass a named source file or route as `--target <path>`. It loads PRODUCT.md, DESIGN.md, the matching surface brief, and native-platform guidance when applicable; follow its directives and do not rerun it.
+1. Run `node <skill-base-dir>/scripts/context.mjs` once per session, where `<skill-base-dir>` is the loaded base directory the runtime reports for this skill; keep cwd at the user's project. That base directory resolves every `node .agents/skills/impeccable/scripts/...` command in this skill and its references, and `.agents/skills/impeccable/scripts` is the fallback only when the runtime reports no base directory. Pass a named source file or route as `--target <path>`. It loads PRODUCT.md, DESIGN.md, the matching surface brief, and native-platform guidance when applicable; follow its directives and do not rerun it.
 2. Before acting, load the one playbook that owns the request: the Commands table's reference for an explicit or clearly implied sub-command, or [reference/new-work.md](reference/new-work.md) for a new surface or replacement visual world. Then inspect the target and at least one representative source of incumbent visual truth (tokens, theme, CSS, component, or asset) before editing.
 3. After analysis and direction are resolved, load [reference/craft-floor.md](reference/craft-floor.md) immediately before editing UI. It carries the quality floor, the absolute bans, and the reflexes no detector catches. Do not load it for planning-only work.
 

.agents/skills/impeccable/agents/impeccable_asset_producer.toml

@@ -13,9 +13,9 @@ Your job is production cleanup, not new art direction. Work only from the approv
 
 Do not redesign. Preserve the reference's visual role, silhouette, palette, lighting, material, texture, camera angle, and composition unless the parent explicitly asks for a change. Preserve perspective only when it belongs to the object or scene itself; if CSS should create the card transform, shadow, rounded clipping, border, or layout, remove that presentation chrome from the raster.
 
-## Decision Sketches
+## Decision Comps
 
-When the parent hands you a decision card packet instead of an approved mock, the job is one sketch: one card, one file, written to the card's declared `sketch` path the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a sketch is reported back, not padded from imagination. Render through the parent's shared frame, including its aspect: the requested surface's first viewport as a flat, matte design sketch in the card's own palette and type character, deliberately unfinished, no photorealism, no gloss; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. The frame is shared across siblings so no sketch looks more finished than another; a finish gap breaks the comparison. The only legible text is the product's real name and one real headline; greek every other text region into indistinct lines, because an invented spec, price, or date in a sketch is a claim PRODUCT.md never made. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a sketch run.
+When the parent hands you a decision card packet instead of an approved mock, the job is one comp: one card, one file, written to the card's declared `sketch` path (the field keeps its wire name) the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a comp is reported back, not padded from imagination. Render the card's direction as a north-star comp at full fidelity: the requested surface's first viewport, prompt led by the surface's own structure (its regions named in order with their scale relationships, never the world's atmosphere), fully committed in the card's own palette, type character, and material world; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. Every sibling renders at the same full fidelity in its own grammar, one surface, one aspect; equal commitment is what keeps the comparison honest. Real product name and real content only; never invent commercial claims, prices, benchmarks, or dates PRODUCT.md does not carry. Write the prompt sidecar beside the file. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a decision-comp run.
 
 ## Input Contract
 
@@ -56,7 +56,7 @@ Ask blockers once, globally. Missing source path/crops or output directory block
 Codex: the imagegen skill's built-in `image_gen` path is the native tool here; prefer it for generation, editing, and the chroma-key workflow.
 7. Remove baked-in UI text, navigation, buttons, body copy, and mock chrome unless the text is part of the asset.
 8. Think through the final DOM/CSS representation before generating. If CSS will own radius, clipping, shadows, borders, perspective, responsive cropping, captions, or card frames, do not bake those into the bitmap.
-9. Save outputs non-destructively in the requested project directory, and leave the intent with the file: after every generation, run `node {{scripts_path}}/embed-prompt.mjs <asset> --prompt "<the prompt used>"` so the prompt is embedded in the image itself, because the build thread composes what you made and needs to know what it is looking at, and the embedding survives copies where sidecars get lost.
+9. Save outputs non-destructively in the requested project directory, and leave the intent with the file: after every generation, run `node .agents/skills/impeccable/scripts/embed-prompt.mjs <asset> --prompt "<the prompt used>"` so the prompt is embedded in the image itself, because the build thread composes what you made and needs to know what it is looking at, and the embedding survives copies where sidecars get lost.
 10. Compare each output against its source crop, opening every image by its workspace-relative path; sandboxed viewers reject absolute paths. If a review/QA tool is available, run it before the final manifest, then retry each major/fatal finding once before finalizing.
 
 Use `texture/pattern extraction` only when the source region is already clean enough to sample as texture. If UI, cards, labels, headings, body copy, or footer chrome must be removed to make a reusable texture or background, classify it as crop-derived cleanup or clean-plate work.

.agents/skills/impeccable/agents/impeccable_finish_reviewer.toml

@@ -13,12 +13,12 @@ A hard turn ceiling ends the run without warning; a run that ends before the fiv
 
 ## Input Contract
 
-Expect: the original request; the confirmed user answers; the artifact path(s); desktop and mobile screenshot paths captured by the parent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and the approved comp path; and the skill's `reference/craft-floor.md` path. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
+Expect: the original request; the confirmed user answers; the artifact path(s); the screenshots the parent captured, which live in `.impeccable/review/` (on the web, `desktop.png` and `mobile.png`; on native, device-class names such as `phone.png` and `tablet.png`, suffixed per OS on adaptive); a screenshot path the calling brief names is authoritative when the file exists, and `.impeccable/review/` is where to look when the brief names none or a named path is missing, never a filename you invent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and, on a comp-led build, the approved comp path (a code-led build has no approved comp; it passes the chosen decision comp as a separate critique-reference input, labeled as such, and nothing in this file that binds “the approved comp” binds it); and the skill's `reference/craft-floor.md` path. On a native (`ios` / `android` / `adaptive`) build the packet also carries the platform reference path(s) (`reference/ios.md` / `reference/android.md`) and a line saying no detector ran: read the platform reference alongside the craft floor and judge every check in the platform's own conventions, the screenshots are device captures rather than browser viewports, and your floor check is the build's only slop gate. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
 
 ## Checks, in order
 
-1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comps with no recorded pick mean the approval point was skipped, and that is a material finding.
-2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
+1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comp-round comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comp-round comps with no recorded pick mean the approval point was skipped, and that is a material finding. Files under `.impeccable/mocks/decision/` are exempt: they are the direction round's dealt hand, produced before any comp round, and they imply no approval whatever the build path; a code-led build has no comp round at all.
+2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. A critique-reference comp, when one arrived on such a build, is provocation rather than spec: no element matrix, no adaptation citations, no asset obligations; its one contribution is the question of what the image dared that the build did not, and the dares worth adopting enter material_fixes as ordinary ordered fixes. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
 3. **Ceiling.** Against the QUALITY BAR card: name the world's native devices the build left unused, frame, depth, lettering treatment, ornament density, motion. The card governs commitment and finish, never composition.
 4. **Contract, promise by promise.** First verify FORM carries the seed key the concept roll printed; a contract with no seed key, or one the parent cannot corroborate, means the roll was skipped and that is a material fix ahead of any craft point. Then, for each of the five blocks, does the render keep the promise? Apply the memory test to the first viewport.
 5. **Truth.** Demonstration data authored and labeled synthetic; no invented commercial claims; unanswered claims present as marked placeholders, not omissions. Every image-native region of the approved comp shipped as a real asset, not a gradient standing in for one, and every produced asset visibly present in the screenshots; an asset applied at near-zero opacity or buried behind other paint is a compliance token, not a shipped material.
@@ -36,5 +36,5 @@ Return the disposition line first, then exactly five sections: `persistence` (pa
 
 ## Verdict Pass
 
-When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.
+When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent recaptures over the same screenshot files you read in the review round, so re-read those exact paths for this round; a round-stamped filename you invent points at nothing. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.
 '''

.agents/skills/impeccable/reference/android.md

@@ -38,3 +38,9 @@ Would a fluent Android user trust this app, or trip on off-spec components? The
 - **One FAB, one primary action.** Never stack FABs or spend one on a secondary task.
 - **Snackbars for transient feedback** (actionable when useful, never a toast for that); dialogs only for decisions that must interrupt.
 - **Material motion patterns.** Container transform, shared-axis, fade-through, with standard easing and durations; honor the system Remove animations setting with a crossfade or instant cut.
+
+## Verifying the build
+
+- **Screenshots come from the emulator or a connected device, never a browser.** Build and install, then capture with `adb exec-out screencap -p > <path>` (pick a device with `adb -s <serial>` when several are attached). Capture every device class the app ships to, at least one phone and, when tablets are a target, one tablet, and write the files where the review flow expects them.
+- **Dark theme and font scale belong in the pass.** `adb shell cmd uimode night yes` flips the theme; `adb shell settings put system font_scale 1.3` (restore `1.0` after) catches the clipped labels a fixed layout hides; with several targets attached, the capture's `-s <serial>` goes on these commands too.
+- **Emulators give breadth; gestures, refresh rates, and performance need hardware.** Say which one produced the evidence.

.agents/skills/impeccable/reference/animate.md

@@ -74,12 +74,15 @@ Keep content visible in the default state so failed scripts do not hide the page
 
 Respect autoplay and sound preferences. Any nonessential loop must stop when offscreen or hidden.
 
+Every web animation needs a `prefers-reduced-motion` path with an intentional alternative. Remove or reduce spatial movement while preserving opacity, color, and state transitions that carry meaning. Reduced motion means fewer and gentler animations, not disabling all motion; feedback that confirms an action should remain legible.
+
 ## Verify
 
 - The focal motion is specific to the selected world and surface.
 - Every supporting animation explains feedback, state, or relationship.
 - Interruption and repeated use behave correctly.
 - Desktop, mobile, and keyboard paths remain usable.
+- The `prefers-reduced-motion` path reduces movement without erasing meaningful feedback or state changes.
 - Expensive effects stay smooth on the target device.
 - Removing an animation would lose meaning or authored character, not merely decoration.
 

.agents/skills/impeccable/reference/bolder.md

@@ -1,5 +1,7 @@
 > **Additional context needed**: which section is the target, and what must stay untouched.
 
+An open direction round owns the word first: "bolder" said while a direction decision is on the table is the Bolder hand register steer, a fresh deal of foreign forms (see new-work.md), not this command. This command refines a surface whose world already shipped.
+
 "Bolder" is an amplification request, and almost always it is scoped to something that already exists. The surrounding page, its system, and its conventions are the given. Your job is to raise one part to the conviction the rest already implies, without rebuilding anything the brief did not name. The reflex answer, reaching for more effects, is the opposite of bold; reject it first.
 
 ## Scope is sovereign

.agents/skills/impeccable/reference/craft-floor.md

@@ -12,6 +12,7 @@ Each of these is a check on the built result, not an intention. Run them togethe
 - **Type:** body measure 65–75ch, display max 6rem, tracking floor -0.04em, balanced headings, obvious scale and weight steps. Run the real copy at every breakpoint and fix what overflows.
 - **Motion:** one authored moment, not scattered effects and not one identical entrance on every section. Exponential ease-out from an already-visible default. Reach past transform and opacity: blur, backdrop-filter, clip-path, mask, and shadow belong to the palette when they stay smooth.
 - **States:** hover, disabled, loading, error, empty. Plus real content, working controls, responsive composition, keyboard focus.
+- **Browser surfaces:** the parts you did not draw still carry the design. Text selection, the caret, custom scrollbars, focus rings, underline offset, and the numerals in tabular data all ship with browser defaults that belong to no design system. Theme them from the palette. This is the cheapest signal that a page was built rather than assembled, and the one models skip most reliably.
 - **Copy:** the product's own language. Controls name their action; errors name the problem and the recovery.
 - **Coverage:** every brief requirement present and findable within seconds.
 

.agents/skills/impeccable/reference/degraded/asset-producer.md

@@ -11,9 +11,9 @@ Your job is production cleanup, not new art direction. Work only from the approv
 
 Do not redesign. Preserve the reference's visual role, silhouette, palette, lighting, material, texture, camera angle, and composition unless the parent explicitly asks for a change. Preserve perspective only when it belongs to the object or scene itself; if CSS should create the card transform, shadow, rounded clipping, border, or layout, remove that presentation chrome from the raster.
 
-## Decision Sketches
+## Decision Comps
 
-When the parent hands you a decision card packet instead of an approved mock, the job is one sketch: one card, one file, written to the card's declared `sketch` path the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a sketch is reported back, not padded from imagination. Render through the parent's shared frame, including its aspect: the requested surface's first viewport as a flat, matte design sketch in the card's own palette and type character, deliberately unfinished, no photorealism, no gloss; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. The frame is shared across siblings so no sketch looks more finished than another; a finish gap breaks the comparison. The only legible text is the product's real name and one real headline; greek every other text region into indistinct lines, because an invented spec, price, or date in a sketch is a claim PRODUCT.md never made. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a sketch run.
+When the parent hands you a decision card packet instead of an approved mock, the job is one comp: one card, one file, written to the card's declared `sketch` path (the field keeps its wire name) the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a comp is reported back, not padded from imagination. Render the card's direction as a north-star comp at full fidelity: the requested surface's first viewport, prompt led by the surface's own structure (its regions named in order with their scale relationships, never the world's atmosphere), fully committed in the card's own palette, type character, and material world; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. Every sibling renders at the same full fidelity in its own grammar, one surface, one aspect; equal commitment is what keeps the comparison honest. Real product name and real content only; never invent commercial claims, prices, benchmarks, or dates PRODUCT.md does not carry. Write the prompt sidecar beside the file. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a decision-comp run.
 
 ## Input Contract
 

.agents/skills/impeccable/reference/degraded/finish-reviewer.md

@@ -11,12 +11,12 @@ A hard turn ceiling ends the run without warning; a run that ends before the fiv
 
 ## Input Contract
 
-Expect: the original request; the confirmed user answers; the artifact path(s); desktop and mobile screenshot paths captured by the parent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and the approved comp path; and the skill's `reference/craft-floor.md` path. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
+Expect: the original request; the confirmed user answers; the artifact path(s); the screenshots the parent captured, which live in `.impeccable/review/` (on the web, `desktop.png` and `mobile.png`; on native, device-class names such as `phone.png` and `tablet.png`, suffixed per OS on adaptive); a screenshot path the calling brief names is authoritative when the file exists, and `.impeccable/review/` is where to look when the brief names none or a named path is missing, never a filename you invent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and, on a comp-led build, the approved comp path (a code-led build has no approved comp; it passes the chosen decision comp as a separate critique-reference input, labeled as such, and nothing in this file that binds “the approved comp” binds it); and the skill's `reference/craft-floor.md` path. On a native (`ios` / `android` / `adaptive`) build the packet also carries the platform reference path(s) (`reference/ios.md` / `reference/android.md`) and a line saying no detector ran: read the platform reference alongside the craft floor and judge every check in the platform's own conventions, the screenshots are device captures rather than browser viewports, and your floor check is the build's only slop gate. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
 
 ## Checks, in order
 
-1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comps with no recorded pick mean the approval point was skipped, and that is a material finding.
-2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
+1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comp-round comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comp-round comps with no recorded pick mean the approval point was skipped, and that is a material finding. Files under `.impeccable/mocks/decision/` are exempt: they are the direction round's dealt hand, produced before any comp round, and they imply no approval whatever the build path; a code-led build has no comp round at all.
+2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. A critique-reference comp, when one arrived on such a build, is provocation rather than spec: no element matrix, no adaptation citations, no asset obligations; its one contribution is the question of what the image dared that the build did not, and the dares worth adopting enter material_fixes as ordinary ordered fixes. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
 3. **Ceiling.** Against the QUALITY BAR card: name the world's native devices the build left unused, frame, depth, lettering treatment, ornament density, motion. The card governs commitment and finish, never composition.
 4. **Contract, promise by promise.** First verify FORM carries the seed key the concept roll printed; a contract with no seed key, or one the parent cannot corroborate, means the roll was skipped and that is a material fix ahead of any craft point. Then, for each of the five blocks, does the render keep the promise? Apply the memory test to the first viewport.
 5. **Truth.** Demonstration data authored and labeled synthetic; no invented commercial claims; unanswered claims present as marked placeholders, not omissions. Every image-native region of the approved comp shipped as a real asset, not a gradient standing in for one, and every produced asset visibly present in the screenshots; an asset applied at near-zero opacity or buried behind other paint is a compliance token, not a shipped material.
@@ -34,4 +34,4 @@ Return the disposition line first, then exactly five sections: `persistence` (pa
 
 ## Verdict Pass
 
-When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.
+When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent recaptures over the same screenshot files you read in the review round, so re-read those exact paths for this round; a round-stamped filename you invent points at nothing. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.

.agents/skills/impeccable/reference/hooks.md

@@ -48,14 +48,20 @@ The first argument is the action. Defaults to `status`.
 5. If `<action>` is `ignore-value`, `ignore-file`, or `ignore-rule`, just print the script output. The default scope is shared `.impeccable/config.json`; add `--local` only when the user explicitly asks for a private exception.
 6. If `<action>` is `status`, just print the script output. Do not add commentary unless the user asked a follow-up question.
 
-## Intentional findings
+## Triage findings
 
-The hook itself never writes ignore config. Persist an exception only after the user explicitly confirms the flagged issue is intentional, and always go through `hook-admin.mjs`.
+The hook itself never writes ignore config; every exception goes through `hook-admin.mjs`. Triage each finding into one of three outcomes:
+
+- **Real design problem**: fix it. Never add an ignore to skip a fix or to push a blocked write through.
+- **Confident false positive or sanctioned exception**: persist the narrowest ignore yourself and disclose it in your reply. The bar is evidence you can name: an intentional demo or fixture, documentation of bad design, literal or domain-appropriate motion (a ball that bounces), or a choice the user already confirmed. Put that evidence in `--reason` as `"<who decided: evidence>"`; write "user confirmed" only when the user actually did.
+- **Unsure**: leave the finding standing and ask the user in one line. Ask once; a one-line question costs less than the hook re-firing on every later edit.
+
+Self-serve stops at `ignore-value`. `ignore-file` and `ignore-rule` silence too much to add on your own judgment; ask the user first.
 
 Prefer the narrowest exception:
 
-- If the finding line shows an exact `ignore-value` command, run that command. This writes shared `.impeccable/config.json` by default.
-- For value-specific findings such as `overused-font` and `bounce-easing`, use `ignore-value` when the user confirms the specific value. Do not use `ignore-rule overused-font` for a specific font.
+- If the finding line shows an `ignore-value <rule> <value>` pair, pass it to `hook-admin.mjs ignore-value` with your `--reason`. This writes shared `.impeccable/config.json` by default.
+- For value-specific findings such as `overused-font` and `bounce-easing`, use `ignore-value` for the specific value. Do not use `ignore-rule overused-font` for a specific font.
 - If the finding has no value-specific command, such as `side-tab`, scope that one rule to the file: `ignore-value <id> "*" --file <path>`. Run `npx impeccable detect <path>` first to see what actually fires there.
 - Reach for `ignore-file <path>` only when the whole file is out of scope for design review: a fixture, a generated artifact, a deliberate slop demo. It silences every rule for that file permanently, including rules that have not been written yet. A real UI surface with one noisy rule wants the file-scoped value ignore above.
 - Use `ignore-rule <id>` only when the user asks to suppress that whole rule across the project. For broad overused-font suppression, use `ignore-rule overused-font --all-values` only when the user asks to ignore overused fonts generally.
@@ -67,10 +73,10 @@ Example value-specific exception:
 node .agents/skills/impeccable/scripts/hook-admin.mjs ignore-value overused-font Inter --shared --reason "User confirmed Inter is intentional"
 ```
 
-Example intentional motion exception:
+Example self-served exception, with the evidence named:
 
 ```bash
-node .agents/skills/impeccable/scripts/hook-admin.mjs ignore-value bounce-easing bounce-ball --shared --reason "User confirmed ball bounce animation is intentional"
+node .agents/skills/impeccable/scripts/hook-admin.mjs ignore-value bounce-easing bounce-ball --shared --reason "Agent: literal ball-bounce animation, bounce easing is the subject"
 ```
 
 Example whole-rule font exception:

.agents/skills/impeccable/reference/ios.md

@@ -43,3 +43,9 @@ Would a fluent iPhone user trust this app, or pause at off-spec controls? The te
 
 - **System transitions.** Push slides, sheets rise, dismiss reverses the entrance. Custom transitions that fight the navigation model disorient.
 - **Honor Reduce Motion.** Crossfade instead of parallax and large slides.
+
+## Verifying the build
+
+- **Screenshots come from the Simulator, never a browser.** Build and run, then capture with `xcrun simctl io booted screenshot <path>` (with several running, replace `booted` with the target's UDID from `xcrun simctl list devices booted`; display names can collide, the UDID never does). Capture every device class the app ships to, at least one iPhone and, when iPad is a target, one iPad, and write the files where the review flow expects them.
+- **Dark Mode and Dynamic Type belong in the pass.** `xcrun simctl ui booted appearance dark` flips appearance, reusing the capture's UDID when several are booted; a check at a large Dynamic Type size catches the truncation a fixed layout hides.
+- **Simulators give breadth; posture, gestures, and performance need hardware.** Say which one produced the evidence.

.agents/skills/impeccable/reference/new-work.md

@@ -43,12 +43,14 @@ The script assigns which structure gets built; your top-ranked structure is what
 1. Name the product's unique mechanism in one sentence, the audience's real scene, its cultural home, and what this first surface must prove. Note the page this category always ships and its predictable opposite; name both as the rut and keep them out of the seven-candidate list. A brief that paints its own picture, a product name, a titled artifact, a governing metaphor, adds its literal reading to the rut: spend at most one candidate on it and derive the rest from elsewhere in the audience's world.
 2. From that cultural world, list seven concrete visual systems, artifacts, places, or rituals the audience knows by heart, each with one line on why it resonates and can carry the mechanism, ordered by resonance. The audience's world includes its graphic and screen traditions, not only its physical objects: the notation, publications, identity programs, data graphics, and interfaces it reads daily; a nameable abstract system (a school of poster, a documentation standard) is as concrete a candidate as any artifact. What would this thing look like as a physical object; what did its world look like before the web? Near-duplicates count once. When more than three of the seven share one material family, the derivation stopped at the subject's most obvious artifact; dig until the list spans at least three families.
 3. Turn that material into complete directions: each joins a reusable visual world to a concrete first-surface experience.
-4. Run `node .agents/skills/impeccable/scripts/concept-seed.mjs --scope direction --mode <mode>` and follow what it prints. This step has no substitute and no skip condition: on a new or replacement world, writing artifact code before this script has run and its assignment is acknowledged is a contract violation, whatever the harness, the model, or the time pressure, because the roll is the mechanism that keeps every run from converging on the category default. The script assigns which direction gets built and deals catalog challengers. Fuse each challenger before judging it: the challenger supplies the form and its system grammar, the product supplies every fact, and clarity wins conflicts. Weigh fused challengers against the assigned direction on exactly two axes, audience identification and product clarity; losing to strong grounded material is a valid outcome, and beating a thin or tool-monoculture list is the point.
-5. Present one direction, fully committed: its world, first viewport, visitor path, signature interaction, cross-surface reach, and honest risk. Alongside it, offer the hand's challengers as named alternates, the weighing's verdict written on each as its one-line case, an honest "fuses poorly because X" included; the weighing informs the user's choice, it never pre-empts it. A hand holds at most three challengers: when the roll deals more, the three strongest join the hand and the rest wait in the re-roll pool, noted in one line; dropping a challenger from the hand itself takes a named product-truth failure, disclosed. Add re-roll with an optional one-line steer. Never present a ranked menu of your own grounded candidates; a lineup of those invites the safest card. The two channels share this structure and differ only in richness: cards and boards on the decision page, names and one-liners through the structured tool; the structured tool's option list also carries the standing exit as its last option.
+4. Run `node .agents/skills/impeccable/scripts/concept-seed.mjs --scope direction --mode <mode>` and follow what it prints. This step has no substitute and no skip condition: on a new or replacement world, writing artifact code before this script has run and its assignment is acknowledged is a contract violation, whatever the harness, the model, or the time pressure, because the roll is the mechanism that keeps every run from converging on the category default. The script assigns which direction gets built and deals catalog challengers. Fuse each challenger before judging it: the challenger supplies the form and its system grammar, the product supplies every fact, and clarity wins conflicts. Weigh fused challengers against the assigned direction on exactly two axes, audience identification and product clarity; losing to strong grounded material is a valid outcome, and beating a thin or tool-monoculture list is the point. The weighing closes with a verdict per challenger, decided before any borrowing is considered: wins (beats the assigned direction on both axes; it becomes the build candidate), competitive (holds one axis; it stays a full alternate), or declined (loses both). A declined challenger is not spent: name the one discipline of its system the assigned direction lacks, and raise the assigned direction to match before presenting it. A donation transfers ambition and system discipline (a palette's total commitment, a grid's density courage, a form's structural honesty), never the challenger's clothes; a motif lifted from a declined world is a costume note, not a raise, and one world owns the page. Write each raise into the presented direction as its own line, named for its donor; a raise nobody can read did not happen.
+5. Present one direction, fully committed and already raised by the hand it beat, its raises visible as named lines: its world, first viewport, visitor path, signature interaction, cross-surface reach, and honest risk. Alongside it, route each dealt challenger by its verdict: winning and competitive challengers are full alternates carrying their QUALITY BAR cards and one-line case, while declined challengers render demoted, compact and quiet, each carrying its verdict plus what the direction kept from it, never full-size and never silently dropped, each still adoptable on request. The verdict informs the user's choice, it never pre-empts it; the demoted row is the hand's proof of judgment, showing why the dealt worlds made the presented direction better. A hand holds at most three full-card challengers: when the roll deals more, the three strongest join the hand and the rest wait in the re-roll pool, noted in one line; dropping a challenger from the hand itself takes a named product-truth failure, disclosed. Add one card for your own top-ranked grounded candidate when it is not the assigned direction, kicker MY PICK, same anatomy as every card, with an honest risk line naming its familiarity when true: the strongest grounded direction is often the one most runs in this category land on, and the user deciding that trade is the point of showing it. Familiar and effective is a legitimate destination, not a failure of nerve; the pick card and the standing exit serve it at two depths. One pick card, never two, never a ranked list: the rest of your grounded candidates stay yours, because a lineup of them hands selection back to a taste function and invites the safest card. The pick never takes the lead position, and when the dice assign your top candidate there is no pick card; the assigned card notes it also topped your list. Add re-roll with an optional one-line steer, offered in three registers: plain (a fresh hand, same spread), safer (the familiar register: your remaining conventional grounded candidates plus the canon against named competitors), and bolder (foreign forms only, at full commitment). A register is the user's steering on the familiar-to-bold axis, never yours to pre-select; when the answer carries one, re-run the seed with `--register <value>` and the next `--reroll` round, and follow what it prints. A user saying "bolder" or "safer" while a direction round is open means these registers, never the bolder or harden commands. The two channels share this structure and differ only in richness: cards and boards on the decision page, names and one-liners through the structured tool; the structured tool's option list carries the assigned direction, the pick, the winning and competitive challengers, and the standing exit as its last option, while declined challengers fold into the assigned option's description as their kept lines, so the raise survives the text channel too.
 
-The standing exit: every direction round offers one quiet, permanent alternative, the category standard, played straight. It is the user's door, never yours: never recommend it, never weigh it against the roll, never let it soften the dealt directions; the counterweights bind the unchosen default, not the chosen one. When the user takes it, in the canon action, a safer-steer, or plain words asking for the familiar or competitor-like path, convention becomes the commitment: ask once for two or three products this should sit alongside, make their craft level the bar, and execute the canon at full fidelity, without irony or smuggled quirk. A standing preference gets recorded as a brand commitment in PRODUCT.md. Re-roll eliminates every direction already shown, grounded and challenger alike; after two consecutive re-rolls, ask what quality is missing. You may re-roll on your own only on named factual grounds, when the assigned direction cannot carry the product's truth or task; taste is never grounds. The user may re-roll freely, and a user- or brief-pinned direction beats the roll, always. Present the decision visually: write an options payload with the assigned direction leading, the dealt challengers as alternates carrying their QUALITY BAR cards, and re-roll, steer, plus canon enabled; a degraded roll with no challengers still uses the page, as a single text-only card with re-roll. Give every card the same anatomy, thesis, palette, materials, first viewport, honest risk, and the challengers' case lines (run the script with `--schema` for the exact shape); the page renders identity from these fields, and a challenger's catalog image rides as labeled inspiration, never as the promise of the build. Author `canonCard` too: the category standard as one honest card with the same anatomy; the page keeps it subordinate, and the counterweights still bind you. Run `node .agents/skills/impeccable/scripts/serve-question.mjs --start --payload <file>` (run it with `--schema` first for the exact payload shape). It daemonizes, prints the page URL and a key, and exits immediately; now open that URL for the user, in-app browser first, then the system opener, then showing the URL. Collect the choice with `--wait --key <key>`, repeating while it exits 3; the ANSWER prints as JSON. Exit 4 means the page was closed without an answer: re-present once through the structured question tool, and with no answer there either, proceed unattended with the assigned direction and state the assumptions. A harness that can leave a shell blocked in the background may instead run the script without `--start` and let it auto-open and block. Only a session where no browser can open at all, headless, CI, an eval worker, a remote shell with no display, puts the same decision through the structured question tool instead; the script self-detects these environments and exits 2 with that advice, so treat exit 2 as this fallback, never as an error to retry.
+The standing exit: every direction round offers one quiet, permanent alternative, the category standard, played straight. It is the user's door, never yours: never recommend it, never weigh it against the roll, never let it soften the dealt directions; the counterweights bind the unchosen default, not the chosen one. When the user takes it, in the canon action, a safer-steer, or plain words asking for the familiar or competitor-like path, convention becomes the commitment: ask once for two or three products this should sit alongside, make their craft level the bar, and execute the canon at full fidelity, without irony or smuggled quirk. A standing preference gets recorded as a brand commitment in PRODUCT.md. Re-roll eliminates every direction already shown, grounded and challenger alike; after two consecutive re-rolls, ask what quality is missing. You may re-roll on your own only on named factual grounds, when the assigned direction cannot carry the product's truth or task; taste is never grounds. The user may re-roll freely, and a user- or brief-pinned direction beats the roll, always. Present the decision visually: write an options payload with the assigned direction leading and its raised lines included, the pick card when one exists, the dealt challengers as alternates carrying their QUALITY BAR cards plus each challenger's verdict and kept line, re-roll with its safer and bolder registers, steer, plus canon enabled, and `followup: true` when the execution-contract round will follow (it does whenever image generation exists and no standing build-path preference is recorded); a degraded roll with no challengers still uses the page, as a single text-only card with re-roll. Give every card the same anatomy, thesis, palette, materials, first viewport, honest risk, and the challengers' case lines (run the script with `--schema` for the exact shape); the page renders identity from these fields, routes declined challengers to a demoted row on its own, and a challenger's catalog image rides as labeled inspiration, never as the promise of the build. Author `canonCard` too: the category standard as one honest card with the same anatomy; the page keeps it subordinate, and the counterweights still bind you. Run `node .agents/skills/impeccable/scripts/serve-question.mjs --start --payload <file>` (run it with `--schema` first for the exact payload shape). It daemonizes, prints the page URL and a key, and exits immediately; now open that URL for the user, in-app browser first, then the system opener, then showing the URL. Collect the choice with `--wait --key <key>`, repeating while it exits 3; the ANSWER prints as JSON. Exit 4 means the page was closed without an answer: re-present once through the structured question tool, and with no answer there either, proceed unattended with the assigned direction and state the assumptions. A harness that can leave a shell blocked in the background may instead run the script without `--start` and let it auto-open and block. Only a session where no browser can open at all, headless, CI, an eval worker, a remote shell with no display, puts the same decision through the structured question tool instead; the script self-detects these environments and exits 2 with that advice, so treat exit 2 as this fallback, never as an error to retry.
 
-When image generation exists, every card also declares a `sketch` path under `.impeccable/sketches/`, the canon card included. Where the harness sandboxes its shell, start the page through the least-sandboxed command path it offers: a sandboxed shell cannot bind the board's port, and the first-attempt failure costs a retry every session. Serve the page first, then produce the sketches; the page shimmer-waits per slot and the user may answer before they land. Render every sketch through one shared frame so the comparison stays about direction, never rendering luck: the requested surface's first viewport as a flat, matte design sketch in that card's own palette and type character, deliberately unfinished, no photorealism, no gloss, identical framing across cards; a candidate whose sketch looks more finished than the others has broken the comparison, not won it. The frame's aspect is the surface's own: a native app or mobile-first surface sketches portrait at its device viewport, a desktop web surface landscape, and the decision page adapts to either, so a phone screen sketched landscape is a broken frame, not a neutral default. The only legible text in a sketch is the product's real name and one real headline; every other text region is greeked, indistinct lines standing where copy will go, because a sketch that renders invented specs, prices, or dates puts claims in front of the user that PRODUCT.md never made. Produce in the order the user reads: the assigned card, then the hand, then canon, each file written the moment it is done. When the harness runs subagents in parallel, fan the set out as one agent per card: each spawn is the shipped asset producer with a single-sketch packet, that card's fields, PRODUCT.md, the shared frame, and the card's declared path, up to four in flight at once. A slot still empty when its agent returns is regenerated inline, and a slot still empty when the user answers is dropped without ceremony; no other supervision is owed. Without parallel subagents, generate in the main thread after serving, in the same reading order, and let the harness's own generation display carry the progress; the wait for the answer follows the last file. A sketch answers which world, never which composition: the comp round still renders its full set, and the chosen card's sketch seeds at most one probe. With no image generation, the cards carry their identity in palette chips and facts, and that page is complete, not a lesser version.
+When image generation exists, every card also declares a `sketch` path under `.impeccable/mocks/decision/` (the field keeps its wire name for compatibility; what it carries is the card's comp), the canon card included. Where the harness sandboxes its shell, start the page through the least-sandboxed command path it offers: a sandboxed shell cannot bind the board's port, and the first-attempt failure costs a retry every session. Serve the page first, then produce the comps; the page shimmer-waits per slot and the user may answer before they land. Each card's image is that direction's north-star comp at full fidelity, produced under the comp discipline in [visualize.md](visualize.md): the requested surface's first viewport, structure-led prompt, real product name and real content, no invented commercial claims, in that card's own palette, type character, and material world, committed all the way. Generation takes the same time at any fidelity, so an unfinished sketch pays sketch quality for comp cost; fairness between cards comes from equal fidelity in each card's own grammar, one surface, one aspect, never from shared unfinishedness. The frame's aspect is the surface's own: a native app or mobile-first surface comps portrait at its device viewport, a desktop web surface landscape, and the decision page adapts to either, so a phone screen comped landscape is a broken frame, not a neutral default. Produce in the order the user reads, the assigned card, then the pick, then the full-card hand, then canon, each file written with its prompt sidecar the moment it is done, so a re-roll's spend front-loads onto the cards read first; declined challengers get no comp, their catalog thumb is their face. When the harness runs subagents in parallel, fan the set out as one agent per card: each spawn is the shipped asset producer with a single-comp packet, that card's fields, PRODUCT.md, the shared frame, and the card's declared path, up to four in flight at once. A slot still empty when its agent returns is regenerated inline, and a slot still empty when the user answers is dropped without ceremony; no other supervision is owed. Without parallel subagents, generate in the main thread after serving, in the same reading order, and let the harness's own generation display carry the progress; the wait for the answer follows the last file. The chosen card's comp is not spent by the choice: on a comp-led build it enters the comp round as compositional option one, and on a code-led build it returns at the finish review as the critique reference, what the image dared that the build did not. The unchosen comps stay in `.impeccable/mocks/decision/` as the round's spent hand; they carry no approval and imply none. With no image generation, the cards carry their identity in palette chips and facts, and that page is complete, not a lesser version; the page then also demotes every challenger's catalog art to a labeled thumbnail on its own, because salience must encode the verdict, never the accident of which cards have images.
+
+The moment the direction lands, one more round on the same open table decides the execution contract. The direction payload declares `followup: true`, so the table stays open after the pick; deliver the build-path payload through `--update` immediately. Two text-only cards. **Comp-led**: a first-viewport comp is generated and it is law, the finish review audits the build against it; boldest composition on the table, fix rounds expected, motion at risk; choosing it makes the comp non-optional, no silent skipping. **Code-led**: no comp of this page and no apology for it; the QUALITY BAR boards still calibrate finish, and the ambition moves into the written contract, the FIRST VIEWPORT block plus a named signature interaction and motion grammar, which the finish reviewer audits in behavior; code-led is not a discount on commitment, the direction still lands fully committed in code. Lead with the chosen world's fit: a costume-heavy catalog world leads comp-led, a quiet or conventional direction leads code-led; the lead is a default, never a decision, and the user flips it freely. A standing preference, voiced once, is recorded as a brand commitment in PRODUCT.md and skips this round on later surfaces. Without image generation there is no fork and no round: code-led is the only path, stated in one line rather than asked. Only a detached table (`--start`) stays open for `--update`: a blocking serve or the structured-tool channel runs the build-path round as its own second question instead, and `followup: true` belongs only on a detached round.
 
 Catalog worlds are working systems, not mood references. When one survives, carry its palette and material, type and composition, topology, controls and state, and responsive rules into the product. When the source is itself an interface language, commit to its native grammar across navigation, content, controls, and states. Open the QUALITY BAR board and hero for the world you build the moment the choice lands, even if you viewed another card earlier; the ANSWER line names the chosen card's images (when the harness only reads files or runs sandboxed, download them into the workspace and open the relative path; sandboxed viewers reject absolute paths outside it). They set the craft level the build must reach, a rendered reference's finish, commitment, and art direction, never the composition; your surface serves this product.
 
@@ -78,7 +80,7 @@ If the work establishes durable strategy for a route or artifact, read its exist
 
 Keep the brief small: scope and visitor mode; audience, job, action/task, proof/content, and constraints; chosen direction and memorable moment; unresolved decisions. Do not copy global product truth or DESIGN.md tokens into it.
 
-Whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports, the locked direction is visualized before it is built, never skipped: load [visualize.md](visualize.md) and follow it, three compositional options rendered and put before the user for approval. This step is proven to produce the most compositional and ambitious work.
+On a comp-led build, whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports, the locked direction is visualized before it is built, never skipped: load [visualize.md](visualize.md) and follow it, three compositional options put before the user for approval, the chosen card's decision comp plus two variations. This step is proven to produce the most compositional and ambitious work. On a code-led build the comp round is skipped by contract, never by drift: the ambition it would have carried lives in the direction contract's FIRST VIEWPORT block and named signature interaction, and the finish reviewer audits those promises in behavior.
 
 For `shape`, return the selected direction to [shape.md](shape.md) and stop before persistence or implementation.
 
@@ -101,8 +103,8 @@ Preserve semantics, accessibility, performance, responsiveness, project conventi
 
 ## 7. Inspect and finish
 
-Inspect desktop and mobile in one batched screenshot round, critique the render against the user's request and the direction contract, fix material gaps, and confirm with one final round; two rounds is the ceiling, and fixes batch between them rather than earning per-tweak screenshots. When an approved comp exists, the critique is a side-by-side: view the comp region and the build region together, the hero and each section as its own crop at legible scale, never one full-page thumbnail, which hides exactly the failures that matter, crude controls, wrong lettering character, flattened material, behind a superficially similar section order. On a Persuade surface, verify the mode did its job: a first-time visitor should know what this is, why it matters, and what to do within seconds, in the form's own vocabulary.
+Inspect the surface's target sizes in one batched screenshot round: desktop and mobile on the web; on a native platform (`ios` / `android` / `adaptive`), the shipped device classes per OS, captured from the simulator or emulator the way the platform reference's Verifying the build section describes. Critique the render against the user's request and the direction contract, fix material gaps, and confirm with one final round; two rounds is the ceiling, and fixes batch between them rather than earning per-tweak screenshots. When an approved comp exists, the critique is a side-by-side: view the comp region and the build region together, the hero and each section as its own crop at legible scale, never one full-page thumbnail, which hides exactly the failures that matter, crude controls, wrong lettering character, flattened material, behind a superficially similar section order. On a Persuade surface, verify the mode did its job: a first-time visitor should know what this is, why it matters, and what to do within seconds, in the form's own vocabulary.
 
-After the second inspection round the build thread's polishing is over: no further defect hunts, micro-edit scripts, or rebuilds here; whatever remains ships through the handoffs, where a fresh context does the finding better and cheaper. Where this harness runs no design hook, run `node .agents/skills/impeccable/scripts/detect.mjs --json` on the changed targets once here, fix what is mechanical, and pass the remaining findings to the reviewer; a hookless build that skips this ships every tell the hook exists to catch. Capture desktop and mobile screenshots to files, then spawn the shipped finish reviewer, `impeccable-finish-reviewer` (`impeccable_finish_reviewer` in codex; `/impeccable-finish-reviewer` in Cursor; on GitHub Copilot say "Use the impeccable-finish-reviewer agent"), with the original request, confirmed answers, the artifact path, the screenshot paths, its direction contract, existing hook findings, the QUALITY BAR card and approved comp paths, and the craft-floor reference path. The reviewer has no browser; screenshots you fail to pass are checks it cannot run. Never read the shipped agents' definition files before spawning; the harness loads them at spawn, and you owe only the input packet. Wait on any agent with one long timeout rather than a loop of short polls, and spend the wait on the next independent step. Verify its return carries the five contract sections; on an empty or thrashed return, respawn once with the same inputs before doing anything else. This review never runs inside the build thread and never inherits it: spawn the reviewer fresh, with no forked conversation history (`fork_turns: 0` in codex); a reviewer that inherits your transcript inherits your framing, your optimism, and your abstractions, and everything it needs travels in the inputs above. Only a harness whose tool surface has no subagent capability at all substitutes a fresh in-thread pass after stepping fully out of the build context, run from [degraded/finish-reviewer.md](degraded/finish-reviewer.md), and a substituted or failed-and-replaced review is disclosed in one line at finish, never silently. When the reviewer's first material fix is a rebuild directive, fidelity failed wholesale rather than in patches, so skip the fix batch and execute the rebuild immediately: re-derive the named regions, produce the named assets, and send the result back for a verdict, telling the user what is happening rather than asking permission to fix a failure. The user is consulted only when a second rebuild directive arrives, both verdicts on the table, or when rebuilding would discard content the user approved. Otherwise apply the material fixes in one batch, rebuild once, and recapture the same viewports. A recapture measures positions, loading, and overflow; it cannot measure whether a fix reached the quality the finding named, so send the recaptured screenshots back to the same reviewer for a verdict scoring every material fix resolved, partial, or unresolved (through the harness's agent continuation; without one, run the scoring fresh from [degraded/finish-reviewer.md](degraded/finish-reviewer.md)'s Verdict Pass). Fixes scored partial or unresolved get another batch, recapture, and verdict. Two rounds is the budget an unattended run ends at; an attended session's ceiling belongs to the user, so when the second verdict still lists open items, put the table in front of them and let them choose between shipping as it stands and funding another round. Whoever is deciding, stop the moment a round resolves nothing, and the reviewer's findings are the only list you work from, never your own re-opened hunt. Report the final verdict table to the user as it stands, open items included, under the reviewer's own disposition word: a table with open material findings is never announced as a pass, and never under a softer label than the reviewer wrote. Do not run a second detector.
+After the second inspection round the build thread's polishing is over: no further defect hunts, micro-edit scripts, or rebuilds here; whatever remains ships through the handoffs, where a fresh context does the finding better and cheaper. On the web, where this harness runs no design hook, run `node .agents/skills/impeccable/scripts/detect.mjs --json` on the changed targets once here, fix what is mechanical, and pass the remaining findings to the reviewer; a hookless web build that skips this ships every tell the hook exists to catch. A native platform skips the detector entirely: it reads HTML and CSS and has no verdict on native code, so the reviewer's floor check is the only slop gate and the input packet says so. Capture the screenshots into `.impeccable/review/`, one file per captured viewport (on the web, `desktop.png` and `mobile.png`; on native, one per device class, such as `phone.png` and `tablet.png`, suffixed per OS on adaptive), creating that directory when the harness does not; the paths you pass the reviewer are its spec, and that directory is where it looks when a passed path is missing. Then spawn the shipped finish reviewer, `impeccable-finish-reviewer` (`impeccable_finish_reviewer` in codex; `/impeccable-finish-reviewer` in Cursor; on GitHub Copilot say "Use the impeccable-finish-reviewer agent"), with the original request, confirmed answers, the artifact path, the screenshot paths, its direction contract, existing hook findings, the QUALITY BAR card and approved comp paths (on a code-led build there is no approved comp; the chosen decision comp rides in that slot as the critique reference, named as such), the craft-floor reference path, and on a native platform the platform reference path(s), [ios.md](ios.md) / [android.md](android.md), both on adaptive, plus one line saying no detector ran, so the reviewer judges in the platform's conventions rather than the web's. The reviewer has no browser; screenshots you fail to pass are checks it cannot run. Never read the shipped agents' definition files before spawning; the harness loads them at spawn, and you owe only the input packet. Wait on any agent with one long timeout rather than a loop of short polls, and spend the wait on the next independent step. Verify its return carries the five contract sections; on an empty or thrashed return, respawn once with the same inputs before doing anything else. This review never runs inside the build thread and never inherits it: spawn the reviewer fresh, with no forked conversation history (`fork_turns: 0` in codex); a reviewer that inherits your transcript inherits your framing, your optimism, and your abstractions, and everything it needs travels in the inputs above. Only a harness whose tool surface has no subagent capability at all substitutes a fresh in-thread pass after stepping fully out of the build context, run from [degraded/finish-reviewer.md](degraded/finish-reviewer.md), and a substituted or failed-and-replaced review is disclosed in one line at finish, never silently. When the reviewer's first material fix is a rebuild directive, fidelity failed wholesale rather than in patches, so skip the fix batch and execute the rebuild immediately: re-derive the named regions, produce the named assets, and send the result back for a verdict, telling the user what is happening rather than asking permission to fix a failure. The user is consulted only when a second rebuild directive arrives, both verdicts on the table, or when rebuilding would discard content the user approved. Otherwise apply the material fixes in one batch, rebuild once, and recapture the same viewports over the same files. A recapture measures positions, loading, and overflow; it cannot measure whether a fix reached the quality the finding named, so send the recaptured screenshots back to the same reviewer for a verdict scoring every material fix resolved, partial, or unresolved (through the harness's agent continuation; without one, run the scoring fresh from [degraded/finish-reviewer.md](degraded/finish-reviewer.md)'s Verdict Pass). Fixes scored partial or unresolved get another batch, recapture, and verdict. Two rounds is the budget an unattended run ends at; an attended session's ceiling belongs to the user, so when the second verdict still lists open items, put the table in front of them and let them choose between shipping as it stands and funding another round. Whoever is deciding, stop the moment a round resolves nothing, and the reviewer's findings are the only list you work from, never your own re-opened hunt. Report the final verdict table to the user as it stands, open items included, under the reviewer's own disposition word: a table with open material findings is never announced as a pass, and never under a softer label than the reviewer wrote. Do not run a second detector.
 
 Then spawn the shipped documenter, `impeccable-documenter` (`impeccable_documenter` in codex), with the project root, the artifact path, the direction contract, PRODUCT.md, the [document.md](document.md) reference path, and the boundary to write at; it records DESIGN.md and the sidecar from the built world, ground truth over intention; without subagents the pass runs from [degraded/documenter.md](degraded/documenter.md). A clean detector pass is not finished; finished is the contract kept, the comp honored, the review closed, and the system recorded.

.agents/skills/impeccable/reference/polish.md

@@ -19,7 +19,7 @@ Fix the cause at the narrowest correct level. Ask when a binding system principl
 
 ## 2. Gather the evidence
 
-Use the feature yourself at representative desktop and mobile sizes. Determine:
+Use the feature yourself at the surface's representative sizes: desktop and mobile on the web; on a native platform (`ios` / `android` / `adaptive`), the shipped device classes on the simulator, emulator, or hardware, captured per the platform reference's Verifying the build section. Determine:
 
 - whether the path is functionally complete;
 - the intended quality bar and time available;
@@ -86,10 +86,10 @@ Do not perfect one corner while leaving the rest below the same quality bar.
 
 Walk the complete path again with mouse, keyboard, and touch where applicable. Check:
 
-- mobile, intermediate, and wide layouts;
+- mobile, intermediate, and wide layouts on the web; phone and tablet size classes in both supported orientations on native;
 - loading, empty, error, success, disabled, long-content, and missing-content states;
 - zoom, contrast, focus, semantics, and screen-reader names;
-- console errors, layout shift, interaction latency, image loading, and supported browsers;
+- console errors, layout shift, interaction latency, and image loading everywhere; supported browsers on the web; supported OS versions, runtime warnings, and dropped frames on native;
 - agreement with DESIGN.md, neighboring features, and the user's scope.
 
 Follow the quality guidance supplied by `context.mjs` and hooks, then run any other relevant QA commands. Context requests a manual scan only when no automatic detector is active; never add another detector pass. Fix real defects and document only narrow intentional exceptions. A clean scan does not replace visual judgment.

.agents/skills/impeccable/reference/visualize.md

@@ -1,12 +1,12 @@
 # Visualize: Direction Comps & Asset Production
 
-Load this from [new-work.md](new-work.md) whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports. PRODUCT.md and DESIGN.md are preconditions. New-work has already resolved the visual world; this file must not reopen it.
+Load this from [new-work.md](new-work.md) on a comp-led build, when image generation is available (a harness-native tool or the API fallback context.mjs reports). A code-led execution contract skips this file by design, not by drift: its ambition lives in the written direction contract and is audited in behavior, so do not load it for a code-led round. PRODUCT.md and DESIGN.md are preconditions. New-work has already resolved the visual world; this file must not reopen it.
 
 The purpose of a probe is to test composition, narrative, hierarchy, density, focal moment, signature use, and image requirements. It is not a second identity workshop. Keep DESIGN.md's palette, typography direction, material language, component character, imagery stance, and motion grammar fixed.
 
 ## Generate three compositional options
 
-Render three distinct high-fidelity north-star comps of the requested surface, with whatever generation capability exists, saved under `.impeccable/mocks/` so they survive the session. Comp at the surface's own viewport: portrait at device size for a native app or mobile-first surface, desktop landscape otherwise; a phone screen comped landscape misstates the composition before anything gets built against it. Comps are the build thread's own work, never delegated: the thread that writes the comp prompts holds the direction's full context, and it has already seen every comp when the build starts. Open every image you produce or reference by its workspace-relative path, never an absolute one: sandboxed viewers reject absolute paths, and everything under the project root has a relative path. Base them on the real content and the surface concepts already developed with the user. Three is the number: one comp invites rubber-stamping, and the spread between three is what surfaces the composition worth building. A decision-page sketch is not a probe: it chose the direction at deliberately unfinished fidelity, so the three comps render regardless, and the chosen card's sketch seeds at most one of them.
+Render three distinct high-fidelity north-star comps of the requested surface, with whatever generation capability exists, saved under `.impeccable/mocks/` so they survive the session. Comp at the surface's own viewport: portrait at device size for a native app or mobile-first surface, desktop landscape otherwise; a phone screen comped landscape misstates the composition before anything gets built against it. Comps are the build thread's own work, never delegated: the thread that writes the comp prompts holds the direction's full context, and it has already seen every comp when the build starts. Open every image you produce or reference by its workspace-relative path, never an absolute one: sandboxed viewers reject absolute paths, and everything under the project root has a relative path. Base them on the real content and the surface concepts already developed with the user. Three is the number: one comp invites rubber-stamping, and the spread between three is what surfaces the composition worth building. The chosen card's decision comp is the first of the three: it already renders this direction at full fidelity under this file's discipline, so this round generates two more that vary what the first held fixed, and all three go to the approval point together. Only a round that arrives with no decision comp, a degraded roll, an identity-mode page, a direction pinned without the decision round, renders all three here.
 
 - A comp is a designed surface, not a picture of the subject. Lead the generation prompt with the surface's own structure, whatever regions this design actually has, named in order with their scale relationships; a page with no navigation states that instead of inventing one, and an unconventional surface states its unconventional skeleton. A prompt that leads with the world's atmosphere gets a vignette back: the model paints the fish market instead of the fish market's website. Self-check every render: if it could hang as a poster, or reads as a photograph or scene with some text on it, it is not a comp; regenerate with the layout scaffold stated more literally.
 - When the user shortlisted multiple concepts, spread the three across them.
@@ -22,7 +22,7 @@ Show the three together: in the harness when it can display images, otherwise on
 
 Do not begin code until the user approves a direction or explicitly delegates the choice. If they delegate, choose using the task brief, PRODUCT.md, and DESIGN.md, and state the evidence. Approval refines the task concept; it does not modify DESIGN.md.
 
-This approval point has no substitute and no skip condition. When the structured question tool errors, fall back to the decision page; only after both fail may you treat the choice as delegated, and a delegated pick is still recorded exactly as an approval is and disclosed in your first reply, not your last. The finish reviewer treats a build with generated comps and no recorded approval as carrying a material finding.
+This approval point has no substitute and no skip condition. When the structured question tool errors, fall back to the decision page; only after both fail may you treat the choice as delegated, and a delegated pick is still recorded exactly as an approval is and disclosed in your first reply, not your last. The finish reviewer treats a build whose comp round produced comps with no recorded approval as carrying a material finding; decision comps under `.impeccable/mocks/decision/` are the direction round's hand, not comp-round output, and imply no approval on their own.
 
 After approval, record the choice where tools can find it: the approved comp's path goes in the surface brief, and the approved comp's `.json` prompt sidecar gains `"approved": true` (every comp generated through `generate-image.mjs` has one; create it if a native tool didn't). The sidecar travels with the mocks folder, so the approval survives sessions and machines that never see the brief. Then summarize the composition and the parts of the comp that must not be literalized, return to new-work.md, record the direction contract from the approved surface concept, and build.
 

.agents/skills/impeccable/scripts/concept-seed.mjs

@@ -31,6 +31,16 @@
  *     recomputes what rounds 0..n-1 drew, excludes all of it, and rolls a
  *     fresh assigned index, challengers, and compositions. One base key therefore
  *     reproduces the entire chain of rounds.
+ *   - REGISTER (--register safer|bolder): the user's steering on the
+ *     familiar-to-bold axis, applied to a re-roll round. A register changes
+ *     only what this round instructs, never what it dealt: the same key and
+ *     reroll count reproduce the same deal whatever the register, so the
+ *     exclusion chain never forks. bolder presents the dealt foreign forms
+ *     as the whole hand (first-dealt leads, dice-assigned by deal order);
+ *     safer spends the dealt hand unseen and presents the familiar register,
+ *     the model's conventional grounded candidates plus the canon against
+ *     named competitors, the one sanctioned lineup of the model's own list.
+ *     Registers are user-requested, never pre-selected by the model.
  *   - RATINGS: the reviewer's approval ratings weight the challenger draw
  *     (3-star doubles the odds, 1-star sits out); the approved pool itself
  *     is unchanged.
@@ -41,7 +51,9 @@
  *   node scripts/concept-seed.mjs --scope surface --mode operate --grain flow
  *   node scripts/concept-seed.mjs --scope direction --candidate-count 6
  *   node scripts/concept-seed.mjs --scope direction --mode persuade --from <key> --reroll 1
- *   node scripts/concept-seed.mjs --chosen <challenger-id> --from <key> --scope direction
+ *   node scripts/concept-seed.mjs --scope direction --mode persuade --from <key> --reroll 1 --register bolder
+ *   node scripts/concept-seed.mjs --chosen <challenger-id> --kind challenger --from <key> --scope direction
+ *   node scripts/concept-seed.mjs --kind assigned --from <key> --scope direction
  *
  * --grain names how much of the product is in play: product, flow, view, or
  * region. A docs site, an onboarding flow, a landing page and a data table are
@@ -62,8 +74,13 @@
  * Challenger data resolves in order: a local catalog directory (the private
  * service repo, evals, and tests set IMPECCABLE_CATALOG_DIR), then the roll
  * API at impeccable.style, then a degraded assignment-only seed when both are
- * unavailable. --chosen sends the anonymous choice ping for API-dealt rolls;
- * DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY disables it.
+ * unavailable. The anonymous choice ping fires once per resolved attended
+ * round on API-dealt rolls: --kind names which card class won (assigned,
+ * pick, challenger, canon) so share metrics have a denominator, --chosen
+ * carries the catalog id when a dealt challenger won, and --register rides
+ * along when the round came from a steered hand. Grounded candidates' names
+ * never leave the machine. DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY disables
+ * the ping entirely.
  *
  * Env vars:
  *   IMPECCABLE_CONCEPT_SEED — same as --from; for reproducible eval runs.
@@ -172,17 +189,35 @@ function telemetryDisabled() {
   return Boolean(process.env.IMPECCABLE_NO_TELEMETRY || process.env.DO_NOT_TRACK);
 }
 
-// Anonymous choice ping: records only that a dealt world was selected.
+// Anonymous choice ping: one per resolved attended direction round. kind
+// says which card class won (assigned / pick / challenger / canon), so
+// pick-share and canon-share have a denominator; chosenId rides along only
+// when a dealt catalog world won, and register only when the round came from
+// a steered hand. Grounded candidates' names never leave the machine: they
+// are derived from the user's project, so the ping carries the kind alone.
 // Fire-and-forget; never fails the caller.
-export async function pingChosen({ chosenId, key, scope, mode }) {
-  if (telemetryDisabled() || !chosenId) return false;
+const PING_KINDS = new Set(['assigned', 'pick', 'challenger', 'canon']);
+export async function pingChosen({ chosenId, key, scope, mode, kind, register }) {
+  if (telemetryDisabled()) return false;
+  if (kind && !PING_KINDS.has(kind)) return false;
+  if (register && register !== 'safer' && register !== 'bolder') return false;
+  // Legacy shape: a bare challenger id with no kind stays a valid ping.
+  if (!chosenId && !kind) return false;
+  if ((kind === 'challenger' || !kind) && !chosenId) return false;
   const controller = new AbortController();
   const timer = setTimeout(() => controller.abort(), apiBudgetMs());
   try {
     await fetch(`${API_BASE}/chosen`, {
       method: 'POST',
       headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify({ chosenId, key, scope, mode }),
+      body: JSON.stringify({
+        ...(chosenId ? { chosenId } : {}),
+        key,
+        scope,
+        mode,
+        ...(kind ? { kind } : {}),
+        ...(register ? { register } : {}),
+      }),
       signal: controller.signal,
     });
     return true;
@@ -260,6 +295,7 @@ export function renderConceptSeed({
   scope = 'surface',
   key = process.env.IMPECCABLE_CONCEPT_SEED || crypto.randomBytes(4).toString('hex'),
   reroll = 0,
+  register = null,
   mode = null,
   grain = null,
   platform = null,
@@ -273,6 +309,15 @@ export function renderConceptSeed({
   if (!Number.isInteger(reroll) || reroll < 0) {
     throw new Error('concept-seed: --reroll must be a non-negative integer');
   }
+  if (register !== null && register !== 'safer' && register !== 'bolder') {
+    throw new Error('concept-seed: --register must be safer or bolder');
+  }
+  if (register !== null && reroll < 1) {
+    throw new Error('concept-seed: --register steers a re-roll round; pass --reroll <n> with it');
+  }
+  if (register !== null && scope !== 'direction') {
+    throw new Error('concept-seed: --register applies to direction rounds only');
+  }
   if (mode !== null && !SEED_MODES.has(mode)) {
     throw new Error('concept-seed: --mode must be persuade, operate, read, or experience');
   }
@@ -326,6 +371,7 @@ export function renderConceptSeed({
         scope,
         key,
         reroll,
+        register,
         mode,
         grain,
         platform,
@@ -357,7 +403,11 @@ export function renderConceptSeed({
   survive the current task plus navigation, quiet and dense content,
   interaction and state, and a substantially different future surface. In an
   attended run, present the assigned direction fully committed and offer
-  re-roll; never present a ranked lineup to choose from. Re-roll yourself only
+  re-roll. You may add ONE card for your top-ranked grounded candidate when
+  it is not the assigned direction, kicker MY PICK, with an honest risk line
+  naming its familiarity; one pick card, never a ranked lineup, and the pick
+  never takes the lead position. When the assignment IS your top candidate,
+  there is no pick card. Re-roll yourself only
   on named factual grounds, when the assignment cannot carry the product's
   truth or task; taste is never grounds.`
   : `After ordering the task's grounded structural candidates by resonance,
@@ -374,7 +424,16 @@ export function renderConceptSeed({
   conflicts. Weigh the fused result against the assigned direction on exactly
   two axes, audience identification and product clarity. Losing to strong
   grounded material is a valid outcome; beating a thin or tool-monoculture
-  list is the point. A fused challenger that wins both axes becomes the build.`
+  list is the point. A fused challenger that wins both axes becomes the build.
+  Close the weighing with a verdict per challenger, decided before any
+  borrowing is considered: wins (beats the assigned direction on both axes),
+  competitive (holds one axis), or declined (loses both). A declined
+  challenger is not spent: name the one discipline of its system the assigned
+  direction lacks, and raise the assigned direction to match before
+  presenting it. A donation transfers ambition and system discipline, never
+  the challenger's clothes; one world owns the page. Write each raise as its
+  own named line on the presented direction, and carry every verdict, kept
+  line, and raise into the decision page payload.`
   : `A challenger wins only when its fused result beats the grounded list on
   audience identification and product clarity. It may change task topology or
   interaction, but never the committed visual identity.`;
@@ -399,8 +458,39 @@ Ambitious motion, spatial media, or interaction is welcome when it strengthens
 the product without weakening semantics, performance, or fallback behavior.`;
 
   if (!data) {
-    return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: degraded; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''} --candidate-count ${candidateCount})
-ASSIGNED INDEX: ${buildIndex}
+    // A degraded roll can still serve the safer register, which needs no
+    // catalog at all: the assignment machinery is suppressed entirely, the
+    // same as the non-degraded safer round, because emitting both "the user
+    // picks" and a mandatory numbered build order hands the model two
+    // contradicting instructions and the mandatory one tends to win. The
+    // bolder register is exactly the thing degradation took away, so it
+    // falls back to a plain grounded round, disclosed.
+    const degradedHeader = `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: degraded; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''}${register ? ` --register ${register}` : ''} --candidate-count ${candidateCount})`;
+    if (register === 'safer') {
+      return `${degradedHeader}
+SAFER REGISTER (user-requested): the assigned index is suspended this
+  round; the user picks, and no candidate is mandated. Present the familiar
+  register: your remaining grounded candidates from the conventional end, at
+  most three, as full cards with an honest risk line each, plus the canon
+  executed against two or three named competitors. This is the one sanctioned
+  lineup of your own ranked candidates; it exists only by this explicit
+  request. When the user voices a standing preference for it, record a brand
+  commitment in PRODUCT.md.
+${authorityInstruction}
+A user- or brief-pinned decision beats the roll, always.
+REGISTER (restated for truncated readers): safer, user-requested; the
+assigned index is suspended this round and the user picks; seed key ${key}.
+`;
+    }
+    const degradedRegister = register === 'bolder'
+      ? `BOLDER REGISTER UNAVAILABLE: bolder deals foreign forms, and this roll ran
+  degraded with no catalog and no roll service, so there is nothing bold to
+  deal. Tell the user, then run this round as a plain grounded re-roll; the
+  assignment below applies.
+`
+      : '';
+    return `${degradedHeader}
+${degradedRegister}ASSIGNED INDEX: ${buildIndex}
   ${promotedInstruction}
   The assignment exists to refuse the model's ranking rut, never to outrank
   the user or the brief. Never expose assignment metadata in user-facing labels.
@@ -471,34 +561,76 @@ structure only, never a palette, typeface, or material. Treat them as serious
 rivals to your habitual layout, and keep only what makes this product clearer.${grainNote}\n`
     : '';
   const rerollBlock = reroll > 0
-    ? `RE-ROLL ROUND ${reroll}: every candidate presented in earlier rounds, grounded
-  and challenger alike, is eliminated and may not return reworded. Derive
+    ? `RE-ROLL ROUND ${reroll}${register ? ` (${register.toUpperCase()} REGISTER, user-requested)` : ''}: every candidate presented in earlier rounds, grounded
+  and challenger alike, is eliminated and may not return reworded.${register ? '' : ` Derive
   genuinely new grounded candidates from unexplored angles before judging
-  these fresh challengers.\n`
+  these fresh challengers.`}\n`
     : '';
+  // A register swaps the round's presentation, never its deal: the assigned
+  // index and challenger fetch stay identical so the chain reproduces, and
+  // only the instructions change.
+  const saferBlock = `SAFER REGISTER: the user asked for the familiar end of the spectrum, so this
+  round's dealt hand is spent unseen, stays excluded from future rounds, and
+  is not printed. The assigned index is suspended this round; the user picks. Present the familiar register: your remaining grounded
+  candidates from the conventional end, at most three, as full cards with an
+  honest risk line each, plus the canon executed against two or three named
+  competitors. This is the one sanctioned lineup of your own ranked
+  candidates; it exists only by this explicit request. When the user voices a
+  standing preference for it, record a brand commitment in PRODUCT.md.`;
+  const bolderBlock = `BOLDER REGISTER: the user asked for foreign forms at full commitment, so no
+  grounded direction is presented this round and the assigned index is
+  suspended. The hand is every dealt challenger below, each fused with the
+  product and presented as a full card; the FIRST dealt challenger leads, an
+  assignment by deal order, so the dice still choose. Verdicts and donations
+  apply between the challengers, weighed against the leader. The pick card
+  sits out; the canon stays, as always.`;
   const telemetryBlock = data.source === 'api'
-    ? `TELEMETRY: if the resolved direction uses one of these challengers, rerun
-  this script once with --chosen <challenger-id> --from ${key} --scope ${scope}${mode ? ` --mode ${mode}` : ''}
-  after resolution. The ping is anonymous (chosen id only) and is skipped
-  automatically when DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY is set.\n`
+    ? `TELEMETRY: after the user's choice resolves, rerun this script once with
+  --kind <assigned|pick|challenger|canon> --from ${key} --scope ${scope}${mode ? ` --mode ${mode}` : ''},
+  adding --chosen <challenger-id> when a dealt challenger won and keeping
+  --register <safer|bolder> when the resolved round came from a steered hand.
+  One ping per resolved attended round. The ping is anonymous, the card kind
+  plus the catalog id when one won; your grounded candidates' names never
+  leave the machine, and the ping is skipped automatically when DO_NOT_TRACK
+  or IMPECCABLE_NO_TELEMETRY is set.\n`
     : '';
-  return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: ${data.source}; approved pool: ${data.poolRevision}; ${data.approvedCount}/${data.catalogCount} human-approved; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''} --candidate-count ${candidateCount} to reproduce this roll against this catalog revision)
-${rerollBlock}ASSIGNED INDEX: ${buildIndex}
+  const assignedBlock = register === null
+    ? `ASSIGNED INDEX: ${buildIndex}
   ${promotedInstruction}
   The assignment exists to refuse the model's ranking rut, never to outrank
-  the user or the brief. Never expose assignment metadata in user-facing labels.
-CHALLENGERS:
+  the user or the brief. Never expose assignment metadata in user-facing labels.`
+    : register === 'safer' ? saferBlock : bolderBlock;
+  // A bolder round has no assigned grounded direction, so the generic
+  // weighing instruction (which measures against the assignment) would
+  // contradict the register; the bolder variant weighs against the leader.
+  const bolderChallengerInstruction = `Fuse each challenger before judging it: the challenger supplies the form
+  and its system grammar, the product supplies every fact, and clarity wins
+  conflicts. Weigh every fused challenger against the fused LEADER, the first
+  dealt, on exactly two axes, audience identification and product clarity;
+  verdicts and donations apply between the challengers, and one that beats
+  the leader on both axes presents as the hand's strongest alternate.`;
+  const roundChallengerInstruction = register === 'bolder' ? bolderChallengerInstruction : challengerInstruction;
+  const challengerSection = register === 'safer'
+    ? ''
+    : `CHALLENGERS:
 ${data.challengers.map(renderChallenger).join('\n')}
-${compositionBlock}${challengerInstruction}
+${compositionBlock}${roundChallengerInstruction}
 When you can view images, open the QUALITY BAR board and hero for any
 challenger you weigh seriously and for the world you build. They exist as a
 craft bar, the finish level and commitment the build is expected to reach,
 never as a mockup to copy; your surface serves this product, not that render.
-${authorityInstruction}
+`;
+  const restated = register === null
+    ? `ASSIGNED INDEX (restated for truncated readers): ${buildIndex}. Build candidate
+${buildIndex} of your own grounded list; seed key ${key}.`
+    : `REGISTER (restated for truncated readers): ${register}, user-requested; the
+assigned index is suspended this round; seed key ${key}.`;
+  return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: ${data.source}; approved pool: ${data.poolRevision}; ${data.approvedCount}/${data.catalogCount} human-approved; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''}${register ? ` --register ${register}` : ''} --candidate-count ${candidateCount} to reproduce this roll against this catalog revision)
+${rerollBlock}${assignedBlock}
+${challengerSection}${authorityInstruction}
 ${richnessInstruction}
 ${telemetryBlock}A user- or brief-pinned decision beats the roll, always.
-ASSIGNED INDEX (restated for truncated readers): ${buildIndex}. Build candidate
-${buildIndex} of your own grounded list; seed key ${key}.
+${restated}
 `;
 }
 
@@ -507,19 +639,25 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
   const fromIdx = args.indexOf('--from');
   const scopeIdx = args.indexOf('--scope');
   const rerollIdx = args.indexOf('--reroll');
+  const registerIdx = args.indexOf('--register');
   const modeIdx = args.indexOf('--mode');
   const grainIdx = args.indexOf('--grain');
   const platformIdx = args.indexOf('--platform');
   const candidateCountIdx = args.indexOf('--candidate-count');
   const chosenIdx = args.indexOf('--chosen');
+  const kindIdx = args.indexOf('--kind');
   try {
-    if (chosenIdx !== -1) {
+    if (chosenIdx !== -1 || kindIdx !== -1) {
       // Choice ping: always exits 0, telemetry must never fail a design flow.
+      // --kind alone pings a non-challenger outcome (assigned/pick/canon);
+      // --chosen alone stays the legacy challenger-win ping.
       const sent = await pingChosen({
-        chosenId: args[chosenIdx + 1],
+        chosenId: chosenIdx !== -1 ? args[chosenIdx + 1] : undefined,
         key: fromIdx !== -1 ? args[fromIdx + 1] : undefined,
         scope: scopeIdx !== -1 ? args[scopeIdx + 1] : undefined,
         mode: modeIdx !== -1 ? args[modeIdx + 1] : undefined,
+        kind: kindIdx !== -1 ? args[kindIdx + 1] : undefined,
+        register: registerIdx !== -1 ? args[registerIdx + 1] : undefined,
       });
       process.stdout.write(sent ? 'choice recorded\n' : 'choice ping skipped\n');
     } else {
@@ -542,6 +680,7 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
           ? args[fromIdx + 1]
           : (process.env.IMPECCABLE_CONCEPT_SEED || crypto.randomBytes(4).toString('hex')),
         reroll: rerollIdx !== -1 ? Number(args[rerollIdx + 1]) : 0,
+        register: registerIdx !== -1 ? args[registerIdx + 1] : null,
         mode: modeIdx !== -1 ? args[modeIdx + 1] : null,
         grain: grainIdx !== -1 ? args[grainIdx + 1] : null,
         platform: platformIdx !== -1 ? args[platformIdx + 1] : null,
@@ -553,6 +692,13 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
     process.exitCode = 1;
   }
   // A raced-out fetch may still hold a socket; exit explicitly so the CLI
-  // never lingers on a dead network path after output is written.
+  // never lingers on a dead network path after output is written. Destroy
+  // fetch's global undici dispatcher first: process.exit() with a live
+  // keep-alive socket trips a libuv assertion on Windows and aborts the
+  // process after a successful roll (nodejs/node#56645).
+  const dispatcher = globalThis[Symbol.for('undici.globalDispatcher.1')];
+  if (dispatcher && typeof dispatcher.destroy === 'function') {
+    try { await dispatcher.destroy(); } catch { /* exit regardless */ }
+  }
   process.exit(process.exitCode ?? 0);
 }

.agents/skills/impeccable/scripts/context-signals.mjs

@@ -22,7 +22,7 @@ import path from 'node:path';
 import { fileURLToPath } from 'node:url';
 import { execFileSync } from 'node:child_process';
 import { loadContext, extractPlatform } from './context.mjs';
-import { getCritiqueDir } from './lib/impeccable-paths.mjs';
+import { readLatestSnapshotAcrossTargets } from './critique-storage.mjs';
 
 /** Is there code here at all, or just context files / an empty repo? */
 function hasCode(cwd) {
@@ -34,23 +34,13 @@ function hasCode(cwd) {
 }
 
 /**
- * The most recent critique snapshot across all targets. Filenames are
- * timestamp-prefixed (`<iso>__<slug>.md`), so a lexical sort is chronological.
- * Parses the small frontmatter for score + P0/P1 counts.
+ * Summarize the most recent critique snapshot across all targets.
  */
 function latestCritique(cwd) {
   try {
-    const dir = getCritiqueDir(cwd);
-    if (!fs.existsSync(dir)) return null;
-    const files = fs.readdirSync(dir).filter((f) => f.endsWith('.md')).sort();
-    if (!files.length) return null;
-    const newest = files[files.length - 1];
-    const text = fs.readFileSync(path.join(dir, newest), 'utf-8');
-    const front = text.split('---')[1] || '';
-    const get = (k) => {
-      const m = front.match(new RegExp(`^${k}:\\s*(.+)$`, 'm'));
-      return m ? m[1].trim() : null;
-    };
+    const latest = readLatestSnapshotAcrossTargets({ cwd });
+    if (!latest) return null;
+    const get = (key) => latest.meta[key] ?? null;
     const num = (v) => {
       const n = Number(v);
       return Number.isFinite(n) ? n : null;
@@ -61,7 +51,7 @@ function latestCritique(cwd) {
       p0: num(get('p0')),
       p1: num(get('p1')),
       timestamp: get('timestamp'),
-      file: path.relative(cwd, path.join(dir, newest)),
+      file: path.relative(cwd, latest.path),
     };
   } catch {
     return null;

.agents/skills/impeccable/scripts/critique-storage.mjs

@@ -105,28 +105,37 @@ function parseFrontmatter(text) {
 }
 
 /**
- * Return all snapshot files for `slug`, sorted oldest → newest.
+ * Return snapshot files matching `suffix`, sorted oldest → newest.
  */
-function listSnapshotsForSlug(slug, cwd) {
+const SNAPSHOT_FILENAME = /^\d{4}-\d{2}-\d{2}T\d{2}-\d{2}-\d{2}Z__.+\.md$/;
+
+function listSnapshots(suffix, cwd) {
   const dir = getCritiqueDir(cwd);
   if (!fs.existsSync(dir)) return [];
-  const suffix = `__${slug}.md`;
   return fs.readdirSync(dir)
-    .filter((f) => f.endsWith(suffix))
+    .filter((f) => SNAPSHOT_FILENAME.test(f) && f.endsWith(suffix))
     .sort()
     .map((f) => path.join(dir, f));
 }
 
+function readLatestSnapshotMatching(suffix, cwd) {
+  const filePath = listSnapshots(suffix, cwd).at(-1);
+  if (!filePath) return null;
+  const body = fs.readFileSync(filePath, 'utf-8');
+  return { path: filePath, body, meta: parseFrontmatter(body) };
+}
+
 /**
  * Return the most recent snapshot for `slug`, or null. Polish reads this
  * to find its fix backlog when the slug matches.
  */
 export function readLatestSnapshot(slug, { cwd = process.cwd() } = {}) {
-  const all = listSnapshotsForSlug(slug, cwd);
-  if (!all.length) return null;
-  const latest = all[all.length - 1];
-  const body = fs.readFileSync(latest, 'utf-8');
-  return { path: latest, body, meta: parseFrontmatter(body) };
+  return readLatestSnapshotMatching(`__${slug}.md`, cwd);
+}
+
+/** Return the most recent snapshot across all targets, or null. */
+export function readLatestSnapshotAcrossTargets({ cwd = process.cwd() } = {}) {
+  return readLatestSnapshotMatching('.md', cwd);
 }
 
 /**
@@ -134,7 +143,7 @@ export function readLatestSnapshot(slug, { cwd = process.cwd() } = {}) {
  * Critique appends a one-line trend to its output using this.
  */
 export function readTrend(slug, { limit = 5, cwd = process.cwd() } = {}) {
-  const all = listSnapshotsForSlug(slug, cwd);
+  const all = listSnapshots(`__${slug}.md`, cwd);
   const slice = all.slice(-limit);
   return slice.map((file) => parseFrontmatter(fs.readFileSync(file, 'utf-8')));
 }

.agents/skills/impeccable/scripts/detector/browser/injected/index.mjs

@@ -683,6 +683,10 @@ if (IS_BROWSER) {
 
       const reasons = collectVisualContrastReasons(el, style);
       if (reasons.length === 0) continue;
+      // Image-only mode filters here, inside the cap: gradient/opacity/filter
+      // candidates earlier in DOM order must not consume the budget and
+      // starve the url()-backed texts this mode exists to sample.
+      if (options.imageOnly && !reasons.includes('image background')) continue;
 
       const textColor = parseRgb(style.color);
       const fontSize = parseFloat(style.fontSize) || 16;
@@ -1175,6 +1179,7 @@ if (IS_BROWSER) {
   }
 
   async function analyzeVisualContrast(options = {}) {
+    // imageOnly is enforced inside the collector, before the candidate cap.
     const candidates = collectVisualContrastCandidates(options);
     const results = [];
     const shouldScrollOffscreen = options.scrollOffscreen === true;
@@ -1260,9 +1265,16 @@ if (IS_BROWSER) {
 
   function addBrowserFindings(groupMap, el, findings) {
     if (!findings || findings.length === 0) return;
+    // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+    // matching findings for its whole subtree. Applied at this choke point so
+    // every per-element attribution (checks, layout, occlusion, rhythm)
+    // honors it; page-level findings attributed to <body> pass through
+    // untouched, since body has no ignoring ancestor.
+    const kept = findings.filter(f => !scopedIgnoreActive(el, f.type));
+    if (kept.length === 0) return;
     const existing = groupMap.get(el);
-    if (existing) existing.push(...findings);
-    else groupMap.set(el, [...findings]);
+    if (existing) existing.push(...kept);
+    else groupMap.set(el, [...kept]);
   }
 
   function browserFindingsFromMap(groupMap) {
@@ -1620,9 +1632,27 @@ if (IS_BROWSER) {
     for (const node of docClone.querySelectorAll('[id^="impeccable-live-"]')) {
       node.remove();
     }
-    const htmlPatternFindings = checkHtmlPatterns(docClone.outerHTML);
-    if (htmlPatternFindings.length > 0) {
-      const mapped = htmlPatternFindings.map(f => {
+    // Regex findings that name a live selector resolve against the real DOM:
+    // pseudo-element/class segments are stripped (the host element is the
+    // anchor), a selector that matches nothing on this page drops the finding
+    // (the CSS ships here, but the pattern never renders — the live DOM is
+    // ground truth in the browser), and a match under a data-impeccable-ignore
+    // ancestor is waived. Selector-less findings stay page-level.
+    const scopedHtmlFindings = checkHtmlPatterns(docClone.outerHTML).filter(f => {
+      if (!f.selector) return true;
+      const query = String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim().replace(/,\s*(?=,|$)/g, '');
+      if (!query || /^[,\s]*$/.test(query)) return true;
+      let matches;
+      try {
+        matches = document.querySelectorAll(query);
+      } catch {
+        return true;
+      }
+      if (matches.length === 0) return false;
+      return [...matches].some(el => !scopedIgnoreActive(el, f.id));
+    });
+    if (scopedHtmlFindings.length > 0) {
+      const mapped = scopedHtmlFindings.map(f => {
         const item = { type: f.id, detail: f.snippet };
         if (f.severity) {
           item.severity = f.severity;
@@ -1652,8 +1682,27 @@ if (IS_BROWSER) {
     };
   }
 
+  // Visual contrast has three modes. Explicit true runs the full sampled
+  // pass; explicit false disables it entirely (the deterministic-only mode
+  // the test suites use). Unset — the default overlay run — samples ONLY
+  // image-backed text: the one class the analytic walk deliberately skips,
+  // because a url() layer's pixels are unknowable without looking. In-page
+  // sampling draws the source image alone to a canvas (glyph ink never
+  // pollutes it), and a cross-origin image without CORS reports unresolved
+  // instead of guessing.
+  function visualContrastMode(options = {}) {
+    const explicit = typeof options.visualContrast === 'boolean'
+      ? options.visualContrast
+      : typeof window.__IMPECCABLE_CONFIG__?.visualContrast === 'boolean'
+        ? window.__IMPECCABLE_CONFIG__.visualContrast
+        : null;
+    if (explicit === true) return 'full';
+    if (explicit === false) return false;
+    return 'image-only';
+  }
+
   function shouldRunVisualContrast(options = {}) {
-    return options.visualContrast === true || window.__IMPECCABLE_CONFIG__?.visualContrast === true;
+    return visualContrastMode(options) !== false;
   }
 
   function visualContrastOptions(options = {}) {
@@ -1830,6 +1879,7 @@ if (IS_BROWSER) {
       return [];
     }
     const resolvedOptions = visualContrastOptions(options);
+    if (visualContrastMode(options) === 'image-only') resolvedOptions.imageOnly = true;
     const analyses = await analyzeVisualContrast(resolvedOptions);
     if (runtime.generation && runtime.generation !== scanGeneration) return analyses;
     lastVisualContrastAnalyses = analyses;

.agents/skills/impeccable/scripts/detector/detect-antipatterns-browser.js

@@ -861,6 +861,34 @@ function checkBorders(tag, widths, colors, radius, opts = {}) {
   return findings;
 }
 
+// ─── Scoped ignores: data-impeccable-ignore ─────────────────────────────────
+//
+// An element-scoped waiver that travels with the markup: any element carrying
+// `data-impeccable-ignore="rule-a rule-b"` (or `*`, or an empty value, for
+// every rule) suppresses matching findings from itself and its entire subtree,
+// in every engine that walks elements — the browser overlay, the extension,
+// and the static scan. This is the DOM twin of the line-based
+// `impeccable-disable` comment directives, which the browser cannot apply (a
+// live DOM has no line numbers), and the generalization of the one-off
+// `data-impeccable-allow-kickers` opt-out.
+//
+// The intended use is curated exhibits: a page that documents anti-patterns by
+// example, or renders a deliberate "before" specimen, marks the container once
+// and every engine skips it while still scanning the page around it.
+function scopedIgnoreActive(el, ruleId) {
+  const rule = String(ruleId || '').toLowerCase();
+  let cur = el;
+  while (cur && cur.nodeType === 1) {
+    const attr = typeof cur.getAttribute === 'function' ? cur.getAttribute('data-impeccable-ignore') : null;
+    if (attr != null) {
+      const rules = String(attr).trim().toLowerCase().split(/[\s,]+/).filter(Boolean);
+      if (rules.length === 0 || rules.includes('*') || rules.includes(rule)) return true;
+    }
+    cur = cur.parentElement;
+  }
+  return false;
+}
+
 // Returns true if the given text is composed entirely of emoji characters
 // (plus whitespace / variation selectors). Emojis render as multicolor glyphs
 // regardless of CSS `color`, so contrast checks against the element's text
@@ -1428,6 +1456,26 @@ function cssTextHasDarkRootBg(content, customProps) {
   return false;
 }
 
+// Best-effort extraction of the CSS selector whose declaration block contains
+// the given index in raw CSS text. Lets CSS-text findings carry a live-DOM
+// anchor, so the browser pass can resolve scoped ignores against the actual
+// element and drop patterns that render nowhere on the page. Returns null for
+// @-rule preludes, keyframe steps, nested blocks, and anything that does not
+// read as a selector; those findings stay page-level.
+function enclosingCssSelector(cssText, index) {
+  if (!cssText || !Number.isFinite(index)) return null;
+  const open = cssText.lastIndexOf('{', index);
+  if (open === -1) return null;
+  const prevClose = Math.max(cssText.lastIndexOf('}', open - 1), cssText.lastIndexOf(';', open - 1));
+  const raw = cssText.slice(prevClose + 1, open).trim().replace(/\s+/g, ' ');
+  if (!raw || raw.startsWith('@') || /^\d/.test(raw) || /[{}<]/.test(raw)) return null;
+  // Keyframe steps: percentage steps fail the digit test above, but `from`
+  // and `to` would read as (never-matching) type selectors and get a valid
+  // finding wrongly dropped by the zero-match rule downstream.
+  if (/^(?:from|to)(?:\s*,\s*(?:from|to))*$/i.test(raw)) return null;
+  return raw;
+}
+
 function scanCssTextForGlow(content) {
   const customProps = collectCssCustomProps(content);
   const hasDarkBg = cssTextHasDarkRootBg(content, customProps);
@@ -1739,6 +1787,7 @@ function scanCssTextForPseudoStripe(rawContent) {
       id: 'side-tab',
       snippet: `${selector} — absolute ${thicknessPx}px pseudo-element stripe (${edge}: 0)`,
       index: selectorStart,
+      selector,
     });
   }
   return findings;
@@ -1801,6 +1850,7 @@ function scanCssTextForInsetStripe(content) {
       findings.push({
         id: 'side-tab',
         snippet: `${selector} — inset box-shadow ${ay === 0 ? ax : ay}px stripe (${edge})`,
+        selector,
       });
       break;
     }
@@ -1858,7 +1908,7 @@ function collectMarqueeKeyframes(content) {
 function scanCssTextForMarquee(content, markup = content) {
   const findings = [];
   if (/<marquee\b/i.test(markup)) {
-    findings.push({ id: 'marquee', snippet: '<marquee> element' });
+    findings.push({ id: 'marquee', snippet: '<marquee> element', selector: 'marquee' });
   }
   const marqueeKeyframes = collectMarqueeKeyframes(content);
   if (marqueeKeyframes.size === 0) return findings;
@@ -1873,7 +1923,7 @@ function scanCssTextForMarquee(content, markup = content) {
       const key = `${selector} ${name}`;
       if (seen.has(key)) continue;
       seen.add(key);
-      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"` });
+      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"`, selector });
     }
   }
   return findings;
@@ -2244,8 +2294,10 @@ function checkHtmlPatterns(html, corpora) {
   const purpleHexRe = /#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9|6366f1|764ba2|667eea)\b/gi;
   if (purpleHexRe.test(styleText)) {
     const purpleTextRe = /(?:(?:^|;)\s*color\s*:\s*(?:.*?)(?:#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9))|gradient.*?#(?:7c3aed|8b5cf6|a855f7|764ba2|667eea))/gi;
-    if (purpleTextRe.test(styleText)) {
-      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected' });
+    purpleTextRe.lastIndex = 0;
+    const purpleMatch = purpleTextRe.exec(styleText);
+    if (purpleMatch) {
+      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected', selector: enclosingCssSelector(styleText, purpleMatch.index + 1) || undefined });
     }
   }
 
@@ -2256,7 +2308,7 @@ function checkHtmlPatterns(html, corpora) {
     const start = Math.max(0, gm.index - 200);
     const context = styleText.substring(start, gm.index + gm[0].length + 200);
     if (/gradient/i.test(context)) {
-      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient' });
+      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient', selector: enclosingCssSelector(styleText, gm.index) || undefined });
       break;
     }
   }
@@ -2322,7 +2374,7 @@ function checkHtmlPatterns(html, corpora) {
     const animationToken = bounceMatch[1]
       .split(/[,\s]+/)
       .find((part) => /bounce|elastic|wobble|jiggle|spring/i.test(part));
-    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}` });
+    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}`, selector: enclosingCssSelector(styleText, bounceMatch.index) || undefined });
   }
 
   // Overshoot cubic-bezier
@@ -2331,7 +2383,7 @@ function checkHtmlPatterns(html, corpora) {
   while ((bm = bezierRe.exec(styleText)) !== null) {
     const y1 = parseFloat(bm[2]), y2 = parseFloat(bm[4]);
     if (y1 < -0.1 || y1 > 1.1 || y2 < -0.1 || y2 > 1.1) {
-      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})` });
+      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})`, selector: enclosingCssSelector(styleText, bm.index) || undefined });
       break;
     }
   }
@@ -2364,18 +2416,21 @@ function checkHtmlPatterns(html, corpora) {
 
   const glowHits = scanCssTextForGlow(styleText);
   if (glowHits.length > 0) {
-    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet });
+    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet, selector: enclosingCssSelector(styleText, glowHits[0].index) || undefined });
   }
 
   // Radial-gradient background halo (gradient-drawn sibling of dark-glow)
   const haloHits = scanCssTextForRadialHalo(styleText);
   if (haloHits.length > 0) {
-    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet });
+    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet, selector: enclosingCssSelector(styleText, haloHits[0].index) || undefined });
   }
 
   // --- Generated-UI tells: repeating-gradient stripes ---
-  if (/repeating-(?:linear|radial|conic)-gradient\s*\(/i.test(styleText)) {
-    findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes' });
+  {
+    const stripesMatch = /repeating-(?:linear|radial|conic)-gradient\s*\(/i.exec(styleText);
+    if (stripesMatch) {
+      findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes', selector: enclosingCssSelector(styleText, stripesMatch.index) || undefined });
+    }
   }
 
   // --- Generated-UI tells: two-axis grid-line background ---
@@ -2393,7 +2448,7 @@ function checkHtmlPatterns(html, corpora) {
   // whole gradient layers.
   const gridHits = scanCssTextForGridBackground(styleText);
   if (gridHits.length > 0) {
-    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet });
+    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet, selector: enclosingCssSelector(styleText, gridHits[0].index) || undefined });
   }
 
   // --- Generated-copy tells: "X theater" framing copy ---
@@ -2413,8 +2468,11 @@ function checkHtmlPatterns(html, corpora) {
   // hover:rotate / hover:translate utility on an <img>. Each distinct
   // mechanism is its own finding.
   const imgHoverCss = /\bimg\b[^,{}]*:hover\b[^{}]*\{[^}]*\btransform\s*:\s*(?:scale|rotate|translate|matrix|skew)/i;
-  if (imgHoverCss.test(styleText)) {
-    findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule' });
+  {
+    const imgHoverMatch = imgHoverCss.exec(styleText);
+    if (imgHoverMatch) {
+      findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule', selector: enclosingCssSelector(styleText, imgHoverMatch.index + imgHoverMatch[0].indexOf('{') + 1) || undefined });
+    }
   }
   const imgTagRe = /<img\b[^>]*\bclass\s*=\s*"([^"]*)"/gi;
   let im;
@@ -2461,6 +2519,33 @@ function readOwnBackgroundColor(el, computedStyle) {
   return bg;
 }
 
+// One element's background-color as the cascade walk sees it: computed style
+// first (with the modern-color fallback), then, in static mode only,
+// custom-prop resolution and the inline-shorthand peek. Shared by
+// resolveBackground and resolveGradientStops so both walks read the same
+// surfaces.
+function readCascadeBackgroundColor(current, style, customPropMap) {
+  let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
+  if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
+    // The static engine can return literal "var(--X)" / "oklch(...)" strings.
+    // Resolve through customPropMap so Tailwind v4 color tokens become RGB.
+    if (customPropMap) {
+      bg = parseColorResolved(style.backgroundColor, customPropMap);
+    }
+    if (!bg || bg.a < 0.1) {
+      // Inline-style fallback for colors the static cascade did not surface
+      // on backgroundColor.
+      const rawStyle = current.getAttribute?.('style') || '';
+      const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
+      const inlineBg = bgMatch ? bgMatch[1].trim() : '';
+      if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
+        bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
+      }
+    }
+  }
+  return bg;
+}
+
 function resolveBackground(el, win, customPropMap) {
   let current = el;
   // Translucent layers (0.1 < a < 1) found on the way down to an opaque
@@ -2489,24 +2574,7 @@ function resolveBackground(el, win, customPropMap) {
     // body backgrounds.
     // Real browsers serialize wide-gamut computed values as oklab()/oklch()
     // (e.g. any color-mix() result), which plain parseRgb misses.
-    let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
-    if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
-      // jsdom returns literal "var(--X)" / "oklch(...)" strings. Resolve
-      // through customPropMap so Tailwind v4 color tokens become RGB.
-      if (customPropMap) {
-        bg = parseColorResolved(style.backgroundColor, customPropMap);
-      }
-      if (!bg || bg.a < 0.1) {
-        // Inline-style fallback. jsdom doesn't decompose background
-        // shorthand, so colors set via inline style are otherwise invisible.
-        const rawStyle = current.getAttribute?.('style') || '';
-        const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
-        const inlineBg = bgMatch ? bgMatch[1].trim() : '';
-        if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
-          bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
-        }
-      }
-    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
 
     if (bg && bg.a > 0.1) {
       if (bg.a >= 0.99) return flatten(bg);
@@ -2524,40 +2592,95 @@ function resolveBackground(el, win, customPropMap) {
     //   • on other elements: bail to null and let the caller fall back
     //     to gradient stops (gradient buttons / hero sections are real
     //     bgs worth checking against).
-    if (hasGradientOrUrl) {
-      if (current.tagName === 'BODY' || current.tagName === 'HTML') {
-        return flatten({ r: 255, g: 255, b: 255, a: 1 });
-      }
-      return null;
-    }
+    // A gradient or image with no solid color under it, at any level
+    // including body/html, means the visible ground is that layer itself.
+    // Return null so the caller measures against the actual gradient stops,
+    // or skips when nothing is parseable — skipping beats a wrong ratio.
+    //
+    // Body/html used to assume white here, a guard written for jsdom, which
+    // never decomposed the `background:` shorthand and so could not see the
+    // solid paper color a texture gradient usually sits on. It turned every
+    // light-on-dark page into a wall of low-contrast false positives (a dark
+    // oklch body gradient produced ~120 "on #ffffff" findings on one site).
+    // Both engines can see shorthand solids now — the browser natively, the
+    // static cascade via expandStaticDeclaration + var() resolution — so a
+    // missing solid is real, and the old failure case cannot recur: opaque
+    // stops fully cover any hidden solid (they ARE the ground), alpha stops
+    // composite over the resolved base or the white canvas default, and
+    // unresolvable stops drop rather than guess.
+    if (hasGradientOrUrl) return null;
     current = current.parentElement;
   }
   return flatten({ r: 255, g: 255, b: 255, a: 1 });
 }
 
+// parseGradientColors (shared) reads only the legacy serializations: rgb()
+// and hex stops. Browsers keep modern-space stops in computed backgroundImage
+// exactly as authored — `linear-gradient(oklch(7% 0.006 95), …)` stays oklch —
+// which is what every token-driven page produces. Route those through
+// parseAnyColor so a gradient ground is measurable rather than invisible.
+function parseGradientColorsModern(bgImage) {
+  if (!bgImage || !/gradient/i.test(bgImage)) return [];
+  const colors = parseGradientColors(bgImage);
+  for (const m of bgImage.matchAll(/(?:oklch|oklab|hsla?|hwb)\(\s*[^()]*\)/gi)) {
+    const c = parseAnyColor(m[0]);
+    if (c) colors.push(c);
+  }
+  return colors;
+}
+
 // Walk parents looking for a gradient background and return its color stops.
 // Used as a fallback when resolveBackground() returns null because the
 // effective background is a gradient (no single solid color to compare against).
+// Translucent solid layers found between the element and the gradient (frosted
+// panels, glass washes) are composited over every stop, the same way
+// resolveBackground flattens them over a solid base — raw stops alone would
+// false-flag dark text on a light frosted wash over a dark gradient, and miss
+// the inverse.
 function resolveGradientStops(el, win, customPropMap) {
   let current = el;
+  const overlays = [];
   while (current && current.nodeType === 1) {
     const style = DETECTOR_IS_BROWSER ? getComputedStyle(current) : win.getComputedStyle(current);
     const bgImage = style.backgroundImage || '';
+    // A url() layer anywhere in the stack — alone, or alongside a gradient in
+    // the same declaration (a translucent wash over a texture photo) — paints
+    // pixels the analytic walk cannot know. Measuring the gradient stops over
+    // the wrong base flagged dark ink sitting on a bright gold-leaf image at
+    // 2.6:1; skipping beats a wrong ratio, and the screenshot subsystem owns
+    // image-backed text.
+    if (bgImage && bgImage !== 'none' && /url\s*\(/i.test(bgImage)) return null;
     let stops = null;
     if (bgImage && bgImage !== 'none' && /gradient/i.test(bgImage)) {
-      const parsed = parseGradientColors(bgImage);
+      const parsed = parseGradientColorsModern(bgImage);
       if (parsed.length > 0) stops = parsed;
     }
     if (!stops && !DETECTOR_IS_BROWSER) {
-      // jsdom doesn't decompose `background:` shorthand — peek at the raw inline style
+      // Static mode: peek at the raw inline style for gradients the cascade did not surface
       const rawStyle = current.getAttribute?.('style') || '';
       const bgMatch = rawStyle.match(/background(?:-image)?\s*:\s*([^;]+)/i);
       if (bgMatch && /gradient/i.test(bgMatch[1])) {
-        const parsed = parseGradientColors(bgMatch[1]);
+        const parsed = parseGradientColorsModern(bgMatch[1]);
         if (parsed.length > 0) stops = parsed;
       }
     }
-    if (stops) return compositeGradientStops(stops, current, win, customPropMap);
+    if (stops) {
+      const composited = compositeGradientStops(stops, current, win, customPropMap);
+      if (!composited || overlays.length === 0) return composited;
+      return composited.map(stop => {
+        let acc = stop;
+        for (let i = overlays.length - 1; i >= 0; i--) acc = compositeColorOver(overlays[i], acc);
+        return acc;
+      });
+    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
+    if (bg && bg.a > 0.1) {
+      // An opaque surface above the gradient means the gradient never shows
+      // through here; resolveBackground would have returned it, so reaching
+      // this is defensive — bail rather than measure the wrong layer.
+      if (bg.a >= 0.99) return null;
+      overlays.push(bg);
+    }
     current = current.parentElement;
   }
   return null;
@@ -2777,6 +2900,10 @@ function checkElementColorsDOM(el) {
   const rect = el.getBoundingClientRect();
   if (rect.width < 10 || rect.height < 10) return [];
   const style = getComputedStyle(el);
+  // Invisible at rest: hidden scene variants (opacity-0 carousels, swap
+  // decks) are not user-visible, and measuring their inherited colors against
+  // whatever surface happens to sit behind the stack is noise, not audit.
+  if (style.visibility === 'hidden' || effectiveOpacityDOM(el) <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
   let effectiveBg = resolveBackground(el);
@@ -3589,11 +3716,13 @@ function checkElementGlowDOM(el) {
   // If resolveBackground returns null (gradient), try to infer from the gradient colors
   let parentBg = el.parentElement ? resolveBackground(el.parentElement) : resolveBackground(el);
   if (!parentBg) {
-    // Gradient background — sample its colors to determine if it's dark
+    // Gradient background — sample its colors to determine if it's dark.
+    // Modern-syntax parsing matters here: body-level gradients now reach this
+    // fallback in browser mode, and their stops usually serialize as oklch.
     let cur = el.parentElement;
     while (cur && cur.nodeType === 1) {
       const bgImage = getComputedStyle(cur).backgroundImage || '';
-      const gradColors = parseGradientColors(bgImage);
+      const gradColors = parseGradientColorsModern(bgImage);
       if (gradColors.length > 0) {
         // Average the gradient colors
         const avg = { r: 0, g: 0, b: 0 };
@@ -4435,6 +4564,14 @@ function checkElementBorders(tag, style, overrides, resolvedRadius, el = null) {
 }
 
 function checkElementColors(el, style, tag, window, customPropMap, hasAnchorInheritRule) {
+  // Invisible at rest, static twin of the browser walk's skip: opacity does
+  // not inherit, so walk ancestors multiplying declared opacity down.
+  if (style.visibility === 'hidden') return [];
+  let effOpacity = 1;
+  for (let cur = el; cur && cur.nodeType === 1 && effOpacity > 0.02; cur = cur.parentElement) {
+    effOpacity *= parseFloat(window.getComputedStyle(cur).opacity || '1');
+  }
+  if (effOpacity <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
 
@@ -5593,6 +5730,11 @@ function isRenderedForBrowserRule(el) {
 function checkElementTextOverflowDOM(el) {
   const tag = el.tagName.toLowerCase();
   if (TEXT_OVERFLOW_SKIP_TAGS.has(tag)) return [];
+  // scrollWidth/clientWidth are CSS box-model metrics; on SVG content Chrome
+  // returns arbitrary non-zero values for both (a <text> reported 78/48 while
+  // its rendered length sat comfortably inside its box), so the delta is
+  // noise, not overflow. SVG clips to its own viewport anyway.
+  if (el.namespaceURI === 'http://www.w3.org/2000/svg') return [];
   if (!isRenderedForBrowserRule(el)) return [];
   // Only the element that actually owns overflowing text — not its ancestors,
   // which inherit a wider scrollWidth from the spilling descendant.
@@ -5977,6 +6119,22 @@ function isPaintedForOcclusion(el) {
 // path is pure geometry and runs anywhere on the page.
 const OCCLUSION_TEXT_SKIP_TAGS = new Set(['script', 'style', 'noscript', 'template', 'title']);
 
+// An element whose effective opacity multiplies out to ~0 paints nothing at
+// rest: it is not user-visible, so visual findings on it (contrast, occlusion)
+// measure a state nobody sees. Browser-only — the walk needs live computed
+// styles. Cycling scenes that fade such elements in later are the screenshot
+// subsystem's territory, not the analytic walk's.
+function effectiveOpacityDOM(el) {
+  let o = 1;
+  // Walk all the way through body and html: `body { opacity: 0 }` page-fade
+  // wrappers hide every descendant just as thoroughly as a local wrapper.
+  for (let cur = el; cur && cur.nodeType === 1; cur = cur.parentElement) {
+    o *= parseFloat(getComputedStyle(cur).opacity || '1');
+    if (o <= 0.02) return 0;
+  }
+  return o;
+}
+
 function checkTextOcclusionDOM() {
   const findings = [];
   const seenVictims = new Set();
@@ -6004,6 +6162,11 @@ function checkTextOcclusionDOM() {
     }
     return false;
   };
+  // The classic occluder shape this rules out is an opacity-0 interaction
+  // layer — a range scrubber stretched over a before/after comparison — which
+  // elementFromPoint still returns and whose UA background-color otherwise
+  // reads as an opaque box.
+  const effectiveOpacity = effectiveOpacityDOM;
 
   // Collect renderable text owners in / near the first viewport for the
   // elementFromPoint probe. SVG <text> counts too.
@@ -6016,6 +6179,7 @@ function checkTextOcclusionDOM() {
     const text = inSvg ? (el.textContent || '').trim() : elementDirectText(el);
     if (text.length < 2) continue;
     if (!isPaintedForOcclusion(el)) continue;
+    if (effectiveOpacity(el) <= 0.02) continue;
     let rect; try { rect = el.getBoundingClientRect(); } catch { continue; }
     if (rect.width < 6 || rect.height < 6) continue;
     // Viewport-bound probe: keep text whose box overlaps the live viewport.
@@ -6049,6 +6213,7 @@ function checkTextOcclusionDOM() {
         if (top === el || el.contains(top) || top.contains(el)) continue;
         const topCs = getComputedStyle(top);
         if (isFloated(topCs) || isMarqueeish(top, topCs) || isPinnedOverlay(top)) continue;
+        if (effectiveOpacity(top) <= 0.02) continue;
         const topTag = top.tagName.toLowerCase();
         // Text sitting under a raw image/video is contrast territory (deduped
         // against the pixel low-contrast rule); leave those alone here.
@@ -6941,6 +7106,10 @@ if (IS_BROWSER) {
 
       const reasons = collectVisualContrastReasons(el, style);
       if (reasons.length === 0) continue;
+      // Image-only mode filters here, inside the cap: gradient/opacity/filter
+      // candidates earlier in DOM order must not consume the budget and
+      // starve the url()-backed texts this mode exists to sample.
+      if (options.imageOnly && !reasons.includes('image background')) continue;
 
       const textColor = parseRgb(style.color);
       const fontSize = parseFloat(style.fontSize) || 16;
@@ -7433,6 +7602,7 @@ if (IS_BROWSER) {
   }
 
   async function analyzeVisualContrast(options = {}) {
+    // imageOnly is enforced inside the collector, before the candidate cap.
     const candidates = collectVisualContrastCandidates(options);
     const results = [];
     const shouldScrollOffscreen = options.scrollOffscreen === true;
@@ -7518,9 +7688,16 @@ if (IS_BROWSER) {
 
   function addBrowserFindings(groupMap, el, findings) {
     if (!findings || findings.length === 0) return;
+    // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+    // matching findings for its whole subtree. Applied at this choke point so
+    // every per-element attribution (checks, layout, occlusion, rhythm)
+    // honors it; page-level findings attributed to <body> pass through
+    // untouched, since body has no ignoring ancestor.
+    const kept = findings.filter(f => !scopedIgnoreActive(el, f.type));
+    if (kept.length === 0) return;
     const existing = groupMap.get(el);
-    if (existing) existing.push(...findings);
-    else groupMap.set(el, [...findings]);
+    if (existing) existing.push(...kept);
+    else groupMap.set(el, [...kept]);
   }
 
   function browserFindingsFromMap(groupMap) {
@@ -7878,9 +8055,27 @@ if (IS_BROWSER) {
     for (const node of docClone.querySelectorAll('[id^="impeccable-live-"]')) {
       node.remove();
     }
-    const htmlPatternFindings = checkHtmlPatterns(docClone.outerHTML);
-    if (htmlPatternFindings.length > 0) {
-      const mapped = htmlPatternFindings.map(f => {
+    // Regex findings that name a live selector resolve against the real DOM:
+    // pseudo-element/class segments are stripped (the host element is the
+    // anchor), a selector that matches nothing on this page drops the finding
+    // (the CSS ships here, but the pattern never renders — the live DOM is
+    // ground truth in the browser), and a match under a data-impeccable-ignore
+    // ancestor is waived. Selector-less findings stay page-level.
+    const scopedHtmlFindings = checkHtmlPatterns(docClone.outerHTML).filter(f => {
+      if (!f.selector) return true;
+      const query = String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim().replace(/,\s*(?=,|$)/g, '');
+      if (!query || /^[,\s]*$/.test(query)) return true;
+      let matches;
+      try {
+        matches = document.querySelectorAll(query);
+      } catch {
+        return true;
+      }
+      if (matches.length === 0) return false;
+      return [...matches].some(el => !scopedIgnoreActive(el, f.id));
+    });
+    if (scopedHtmlFindings.length > 0) {
+      const mapped = scopedHtmlFindings.map(f => {
         const item = { type: f.id, detail: f.snippet };
         if (f.severity) {
           item.severity = f.severity;
@@ -7910,8 +8105,27 @@ if (IS_BROWSER) {
     };
   }
 
+  // Visual contrast has three modes. Explicit true runs the full sampled
+  // pass; explicit false disables it entirely (the deterministic-only mode
+  // the test suites use). Unset — the default overlay run — samples ONLY
+  // image-backed text: the one class the analytic walk deliberately skips,
+  // because a url() layer's pixels are unknowable without looking. In-page
+  // sampling draws the source image alone to a canvas (glyph ink never
+  // pollutes it), and a cross-origin image without CORS reports unresolved
+  // instead of guessing.
+  function visualContrastMode(options = {}) {
+    const explicit = typeof options.visualContrast === 'boolean'
+      ? options.visualContrast
+      : typeof window.__IMPECCABLE_CONFIG__?.visualContrast === 'boolean'
+        ? window.__IMPECCABLE_CONFIG__.visualContrast
+        : null;
+    if (explicit === true) return 'full';
+    if (explicit === false) return false;
+    return 'image-only';
+  }
+
   function shouldRunVisualContrast(options = {}) {
-    return options.visualContrast === true || window.__IMPECCABLE_CONFIG__?.visualContrast === true;
+    return visualContrastMode(options) !== false;
   }
 
   function visualContrastOptions(options = {}) {
@@ -8088,6 +8302,7 @@ if (IS_BROWSER) {
       return [];
     }
     const resolvedOptions = visualContrastOptions(options);
+    if (visualContrastMode(options) === 'image-only') resolvedOptions.imageOnly = true;
     const analyses = await analyzeVisualContrast(resolvedOptions);
     if (runtime.generation && runtime.generation !== scanGeneration) return analyses;
     lastVisualContrastAnalyses = analyses;

.agents/skills/impeccable/scripts/detector/detect-antipatterns.mjs

@@ -35,6 +35,7 @@ export { detectUrl, createBrowserDetector } from './engines/browser/detect-url.m
 export { detectText, extractStyleBlocks, extractCSSinJS } from './engines/regex/detect-text.mjs';
 export {
   walkDir,
+  hasScannableExtension,
   SCANNABLE_EXTENSIONS,
   SKIP_DIRS,
   buildImportGraph,

.agents/skills/impeccable/scripts/detector/engines/static-html/css-cascade.mjs

@@ -226,6 +226,10 @@ const STATIC_INHERITED_PROPS = new Set([
   'color', 'fontFamily', 'fontSize', 'fontStyle', 'fontWeight', 'fontVariant',
   'lineHeight', 'letterSpacing', 'textTransform', 'textAlign', 'hyphens',
   'webkitHyphens',
+  // visibility inherits in real CSS, and the invisible-at-rest contrast skip
+  // relies on descendants of a hidden container computing as hidden. A child
+  // that declares `visibility: visible` still overrides the inherited value.
+  'visibility',
 ]);
 
 const STATIC_DEFAULT_STYLE = {
@@ -278,6 +282,7 @@ const STATIC_DEFAULT_STYLE = {
   marginLeft: '0px',
   position: 'static',
   visibility: 'visible',
+  opacity: '1',
   top: 'auto',
   right: 'auto',
   bottom: 'auto',
@@ -334,6 +339,7 @@ const STATIC_PROP_MAP = {
   'margin-left': 'marginLeft',
   'position': 'position',
   'visibility': 'visibility',
+  'opacity': 'opacity',
   'top': 'top',
   'right': 'right',
   'bottom': 'bottom',

.agents/skills/impeccable/scripts/detector/engines/static-html/detect-html.mjs

@@ -28,6 +28,7 @@ import {
   checkCreamPalette,
   checkHtmlPatterns,
   checkKickerAboveHeadingFromDoc,
+  scopedIgnoreActive,
   checkNumberedSectionLabelsFromDoc,
   checkPageLayout,
   checkPageQualityFromDoc,
@@ -138,10 +139,21 @@ async function detectHtml(filePath, options = {}) {
         domutils,
       };
     });
-  } catch {
-    return detectText(html, filePath, options);
+  } catch (err) {
+  if (!globalThis.__impeccableStaticHtmlWarned) {
+    globalThis.__impeccableStaticHtmlWarned = true;
+
+    process.stderr.write(
+  'impeccable detect: DEGRADED - HTML parser modules unavailable ' +
+  '(htmlparser2, css-select, css-tree, domutils).\n' +
+  'Falling back to regex matching. Custom properties, selector matching and computed ' +
+  'contrast are NOT evaluated; findings are an undercount, not a clean bill of health.\n'
+);
   }
 
+  return detectText(html, filePath, options);
+}
+
   const resolvedPath = path.resolve(filePath);
   const fileDir = path.dirname(resolvedPath);
   const root = profileStep(profile, {
@@ -171,6 +183,9 @@ async function detectHtml(filePath, options = {}) {
       const tag = el.tagName.toLowerCase();
       const style = window.getComputedStyle(el);
       for (const f of runElementCheck(rule.id, () => rule.run(el, tag, style, window, customPropMap))) {
+        // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+        // matching findings for its subtree, same as the browser walk.
+        if (scopedIgnoreActive(el, f.id)) continue;
         findings.push(finding(f.id, filePath, f.snippet));
       }
     }
@@ -238,6 +253,17 @@ async function detectHtml(filePath, options = {}) {
     for (const f of runPageCheck('html-patterns', () => checkHtmlPatterns(html, patternCorpora).filter(item =>
       item.id !== 'bounce-easing' && item.id !== 'layout-transition'
     ))) {
+      // Selector-backed page findings honor scoped waivers here too, matching
+      // the browser pass: resolve the selector and drop the finding when an
+      // ignoring ancestor covers a match. Unlike the browser, an unmatched
+      // selector keeps the finding — static scans see partial documents.
+      if (f.selector) {
+        let matches = null;
+        try {
+          matches = document.querySelectorAll(String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim());
+        } catch { matches = null; }
+        if (matches && matches.length > 0 && [...matches].every(el => scopedIgnoreActive(el, f.id))) continue;
+      }
       const item = finding(f.id, filePath, f.snippet);
       // Position-aware severity promotion: checks may attach a per-finding
       // severity (e.g. a pulsing dot inside a header/nav landmark) that

.agents/skills/impeccable/scripts/detector/node/file-system.mjs

@@ -26,11 +26,20 @@ const HIDDEN_SOURCE_DIRS = new Set(['.vitepress', '.vuepress', '.storybook']);
 const SCANNABLE_EXTENSIONS = new Set([
   '.html', '.htm', '.css', '.scss', '.sass', '.less',
   '.jsx', '.tsx', '.js', '.ts',
-  '.vue', '.svelte', '.astro',
+  '.vue', '.svelte', '.astro', '.blade.php',
 ]);
 
 const HTML_EXTENSIONS = new Set(['.html', '.htm']);
 
+function hasScannableExtension(filename) {
+  const lower = filename.toLowerCase();
+  if (SCANNABLE_EXTENSIONS.has(path.extname(lower))) return true;
+  for (const ext of SCANNABLE_EXTENSIONS) {
+    if (ext.indexOf('.', 1) !== -1 && lower.endsWith(ext)) return true;
+  }
+  return false;
+}
+
 const IMPORT_SPECIFIER_PATTERNS = [
   /import\s+(?:[\s\S]*?from\s+)?['"]([^'"]+)['"]/g,
   /@import\s+(?:url\(\s*)?['"]?([^'");\s]+)['"]?\s*\)?/g,
@@ -46,7 +55,7 @@ function walkDir(dir) {
     if (entry.isDirectory() && entry.name.startsWith('.') && !HIDDEN_SOURCE_DIRS.has(entry.name)) continue;
     const full = path.join(dir, entry.name);
     if (entry.isDirectory()) files.push(...walkDir(full));
-    else if (SCANNABLE_EXTENSIONS.has(path.extname(entry.name).toLowerCase())) files.push(full);
+    else if (hasScannableExtension(entry.name)) files.push(full);
   }
   return files;
 }
@@ -194,6 +203,7 @@ export {
   SKIP_DIRS,
   SCANNABLE_EXTENSIONS,
   HTML_EXTENSIONS,
+  hasScannableExtension,
   walkDir,
   resolveImport,
   buildImportGraph,

.agents/skills/impeccable/scripts/detector/rules/checks.mjs

@@ -70,6 +70,34 @@ function checkBorders(tag, widths, colors, radius, opts = {}) {
   return findings;
 }
 
+// ─── Scoped ignores: data-impeccable-ignore ─────────────────────────────────
+//
+// An element-scoped waiver that travels with the markup: any element carrying
+// `data-impeccable-ignore="rule-a rule-b"` (or `*`, or an empty value, for
+// every rule) suppresses matching findings from itself and its entire subtree,
+// in every engine that walks elements — the browser overlay, the extension,
+// and the static scan. This is the DOM twin of the line-based
+// `impeccable-disable` comment directives, which the browser cannot apply (a
+// live DOM has no line numbers), and the generalization of the one-off
+// `data-impeccable-allow-kickers` opt-out.
+//
+// The intended use is curated exhibits: a page that documents anti-patterns by
+// example, or renders a deliberate "before" specimen, marks the container once
+// and every engine skips it while still scanning the page around it.
+function scopedIgnoreActive(el, ruleId) {
+  const rule = String(ruleId || '').toLowerCase();
+  let cur = el;
+  while (cur && cur.nodeType === 1) {
+    const attr = typeof cur.getAttribute === 'function' ? cur.getAttribute('data-impeccable-ignore') : null;
+    if (attr != null) {
+      const rules = String(attr).trim().toLowerCase().split(/[\s,]+/).filter(Boolean);
+      if (rules.length === 0 || rules.includes('*') || rules.includes(rule)) return true;
+    }
+    cur = cur.parentElement;
+  }
+  return false;
+}
+
 // Returns true if the given text is composed entirely of emoji characters
 // (plus whitespace / variation selectors). Emojis render as multicolor glyphs
 // regardless of CSS `color`, so contrast checks against the element's text
@@ -637,6 +665,26 @@ function cssTextHasDarkRootBg(content, customProps) {
   return false;
 }
 
+// Best-effort extraction of the CSS selector whose declaration block contains
+// the given index in raw CSS text. Lets CSS-text findings carry a live-DOM
+// anchor, so the browser pass can resolve scoped ignores against the actual
+// element and drop patterns that render nowhere on the page. Returns null for
+// @-rule preludes, keyframe steps, nested blocks, and anything that does not
+// read as a selector; those findings stay page-level.
+function enclosingCssSelector(cssText, index) {
+  if (!cssText || !Number.isFinite(index)) return null;
+  const open = cssText.lastIndexOf('{', index);
+  if (open === -1) return null;
+  const prevClose = Math.max(cssText.lastIndexOf('}', open - 1), cssText.lastIndexOf(';', open - 1));
+  const raw = cssText.slice(prevClose + 1, open).trim().replace(/\s+/g, ' ');
+  if (!raw || raw.startsWith('@') || /^\d/.test(raw) || /[{}<]/.test(raw)) return null;
+  // Keyframe steps: percentage steps fail the digit test above, but `from`
+  // and `to` would read as (never-matching) type selectors and get a valid
+  // finding wrongly dropped by the zero-match rule downstream.
+  if (/^(?:from|to)(?:\s*,\s*(?:from|to))*$/i.test(raw)) return null;
+  return raw;
+}
+
 function scanCssTextForGlow(content) {
   const customProps = collectCssCustomProps(content);
   const hasDarkBg = cssTextHasDarkRootBg(content, customProps);
@@ -948,6 +996,7 @@ function scanCssTextForPseudoStripe(rawContent) {
       id: 'side-tab',
       snippet: `${selector} — absolute ${thicknessPx}px pseudo-element stripe (${edge}: 0)`,
       index: selectorStart,
+      selector,
     });
   }
   return findings;
@@ -1010,6 +1059,7 @@ function scanCssTextForInsetStripe(content) {
       findings.push({
         id: 'side-tab',
         snippet: `${selector} — inset box-shadow ${ay === 0 ? ax : ay}px stripe (${edge})`,
+        selector,
       });
       break;
     }
@@ -1067,7 +1117,7 @@ function collectMarqueeKeyframes(content) {
 function scanCssTextForMarquee(content, markup = content) {
   const findings = [];
   if (/<marquee\b/i.test(markup)) {
-    findings.push({ id: 'marquee', snippet: '<marquee> element' });
+    findings.push({ id: 'marquee', snippet: '<marquee> element', selector: 'marquee' });
   }
   const marqueeKeyframes = collectMarqueeKeyframes(content);
   if (marqueeKeyframes.size === 0) return findings;
@@ -1082,7 +1132,7 @@ function scanCssTextForMarquee(content, markup = content) {
       const key = `${selector} ${name}`;
       if (seen.has(key)) continue;
       seen.add(key);
-      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"` });
+      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"`, selector });
     }
   }
   return findings;
@@ -1453,8 +1503,10 @@ function checkHtmlPatterns(html, corpora) {
   const purpleHexRe = /#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9|6366f1|764ba2|667eea)\b/gi;
   if (purpleHexRe.test(styleText)) {
     const purpleTextRe = /(?:(?:^|;)\s*color\s*:\s*(?:.*?)(?:#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9))|gradient.*?#(?:7c3aed|8b5cf6|a855f7|764ba2|667eea))/gi;
-    if (purpleTextRe.test(styleText)) {
-      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected' });
+    purpleTextRe.lastIndex = 0;
+    const purpleMatch = purpleTextRe.exec(styleText);
+    if (purpleMatch) {
+      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected', selector: enclosingCssSelector(styleText, purpleMatch.index + 1) || undefined });
     }
   }
 
@@ -1465,7 +1517,7 @@ function checkHtmlPatterns(html, corpora) {
     const start = Math.max(0, gm.index - 200);
     const context = styleText.substring(start, gm.index + gm[0].length + 200);
     if (/gradient/i.test(context)) {
-      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient' });
+      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient', selector: enclosingCssSelector(styleText, gm.index) || undefined });
       break;
     }
   }
@@ -1531,7 +1583,7 @@ function checkHtmlPatterns(html, corpora) {
     const animationToken = bounceMatch[1]
       .split(/[,\s]+/)
       .find((part) => /bounce|elastic|wobble|jiggle|spring/i.test(part));
-    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}` });
+    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}`, selector: enclosingCssSelector(styleText, bounceMatch.index) || undefined });
   }
 
   // Overshoot cubic-bezier
@@ -1540,7 +1592,7 @@ function checkHtmlPatterns(html, corpora) {
   while ((bm = bezierRe.exec(styleText)) !== null) {
     const y1 = parseFloat(bm[2]), y2 = parseFloat(bm[4]);
     if (y1 < -0.1 || y1 > 1.1 || y2 < -0.1 || y2 > 1.1) {
-      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})` });
+      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})`, selector: enclosingCssSelector(styleText, bm.index) || undefined });
       break;
     }
   }
@@ -1573,18 +1625,21 @@ function checkHtmlPatterns(html, corpora) {
 
   const glowHits = scanCssTextForGlow(styleText);
   if (glowHits.length > 0) {
-    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet });
+    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet, selector: enclosingCssSelector(styleText, glowHits[0].index) || undefined });
   }
 
   // Radial-gradient background halo (gradient-drawn sibling of dark-glow)
   const haloHits = scanCssTextForRadialHalo(styleText);
   if (haloHits.length > 0) {
-    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet });
+    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet, selector: enclosingCssSelector(styleText, haloHits[0].index) || undefined });
   }
 
   // --- Generated-UI tells: repeating-gradient stripes ---
-  if (/repeating-(?:linear|radial|conic)-gradient\s*\(/i.test(styleText)) {
-    findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes' });
+  {
+    const stripesMatch = /repeating-(?:linear|radial|conic)-gradient\s*\(/i.exec(styleText);
+    if (stripesMatch) {
+      findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes', selector: enclosingCssSelector(styleText, stripesMatch.index) || undefined });
+    }
   }
 
   // --- Generated-UI tells: two-axis grid-line background ---
@@ -1602,7 +1657,7 @@ function checkHtmlPatterns(html, corpora) {
   // whole gradient layers.
   const gridHits = scanCssTextForGridBackground(styleText);
   if (gridHits.length > 0) {
-    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet });
+    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet, selector: enclosingCssSelector(styleText, gridHits[0].index) || undefined });
   }
 
   // --- Generated-copy tells: "X theater" framing copy ---
@@ -1622,8 +1677,11 @@ function checkHtmlPatterns(html, corpora) {
   // hover:rotate / hover:translate utility on an <img>. Each distinct
   // mechanism is its own finding.
   const imgHoverCss = /\bimg\b[^,{}]*:hover\b[^{}]*\{[^}]*\btransform\s*:\s*(?:scale|rotate|translate|matrix|skew)/i;
-  if (imgHoverCss.test(styleText)) {
-    findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule' });
+  {
+    const imgHoverMatch = imgHoverCss.exec(styleText);
+    if (imgHoverMatch) {
+      findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule', selector: enclosingCssSelector(styleText, imgHoverMatch.index + imgHoverMatch[0].indexOf('{') + 1) || undefined });
+    }
   }
   const imgTagRe = /<img\b[^>]*\bclass\s*=\s*"([^"]*)"/gi;
   let im;
@@ -1670,6 +1728,33 @@ function readOwnBackgroundColor(el, computedStyle) {
   return bg;
 }
 
+// One element's background-color as the cascade walk sees it: computed style
+// first (with the modern-color fallback), then, in static mode only,
+// custom-prop resolution and the inline-shorthand peek. Shared by
+// resolveBackground and resolveGradientStops so both walks read the same
+// surfaces.
+function readCascadeBackgroundColor(current, style, customPropMap) {
+  let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
+  if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
+    // The static engine can return literal "var(--X)" / "oklch(...)" strings.
+    // Resolve through customPropMap so Tailwind v4 color tokens become RGB.
+    if (customPropMap) {
+      bg = parseColorResolved(style.backgroundColor, customPropMap);
+    }
+    if (!bg || bg.a < 0.1) {
+      // Inline-style fallback for colors the static cascade did not surface
+      // on backgroundColor.
+      const rawStyle = current.getAttribute?.('style') || '';
+      const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
+      const inlineBg = bgMatch ? bgMatch[1].trim() : '';
+      if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
+        bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
+      }
+    }
+  }
+  return bg;
+}
+
 function resolveBackground(el, win, customPropMap) {
   let current = el;
   // Translucent layers (0.1 < a < 1) found on the way down to an opaque
@@ -1698,24 +1783,7 @@ function resolveBackground(el, win, customPropMap) {
     // body backgrounds.
     // Real browsers serialize wide-gamut computed values as oklab()/oklch()
     // (e.g. any color-mix() result), which plain parseRgb misses.
-    let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
-    if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
-      // jsdom returns literal "var(--X)" / "oklch(...)" strings. Resolve
-      // through customPropMap so Tailwind v4 color tokens become RGB.
-      if (customPropMap) {
-        bg = parseColorResolved(style.backgroundColor, customPropMap);
-      }
-      if (!bg || bg.a < 0.1) {
-        // Inline-style fallback. jsdom doesn't decompose background
-        // shorthand, so colors set via inline style are otherwise invisible.
-        const rawStyle = current.getAttribute?.('style') || '';
-        const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
-        const inlineBg = bgMatch ? bgMatch[1].trim() : '';
-        if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
-          bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
-        }
-      }
-    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
 
     if (bg && bg.a > 0.1) {
       if (bg.a >= 0.99) return flatten(bg);
@@ -1733,40 +1801,95 @@ function resolveBackground(el, win, customPropMap) {
     //   • on other elements: bail to null and let the caller fall back
     //     to gradient stops (gradient buttons / hero sections are real
     //     bgs worth checking against).
-    if (hasGradientOrUrl) {
-      if (current.tagName === 'BODY' || current.tagName === 'HTML') {
-        return flatten({ r: 255, g: 255, b: 255, a: 1 });
-      }
-      return null;
-    }
+    // A gradient or image with no solid color under it, at any level
+    // including body/html, means the visible ground is that layer itself.
+    // Return null so the caller measures against the actual gradient stops,
+    // or skips when nothing is parseable — skipping beats a wrong ratio.
+    //
+    // Body/html used to assume white here, a guard written for jsdom, which
+    // never decomposed the `background:` shorthand and so could not see the
+    // solid paper color a texture gradient usually sits on. It turned every
+    // light-on-dark page into a wall of low-contrast false positives (a dark
+    // oklch body gradient produced ~120 "on #ffffff" findings on one site).
+    // Both engines can see shorthand solids now — the browser natively, the
+    // static cascade via expandStaticDeclaration + var() resolution — so a
+    // missing solid is real, and the old failure case cannot recur: opaque
+    // stops fully cover any hidden solid (they ARE the ground), alpha stops
+    // composite over the resolved base or the white canvas default, and
+    // unresolvable stops drop rather than guess.
+    if (hasGradientOrUrl) return null;
     current = current.parentElement;
   }
   return flatten({ r: 255, g: 255, b: 255, a: 1 });
 }
 
+// parseGradientColors (shared) reads only the legacy serializations: rgb()
+// and hex stops. Browsers keep modern-space stops in computed backgroundImage
+// exactly as authored — `linear-gradient(oklch(7% 0.006 95), …)` stays oklch —
+// which is what every token-driven page produces. Route those through
+// parseAnyColor so a gradient ground is measurable rather than invisible.
+function parseGradientColorsModern(bgImage) {
+  if (!bgImage || !/gradient/i.test(bgImage)) return [];
+  const colors = parseGradientColors(bgImage);
+  for (const m of bgImage.matchAll(/(?:oklch|oklab|hsla?|hwb)\(\s*[^()]*\)/gi)) {
+    const c = parseAnyColor(m[0]);
+    if (c) colors.push(c);
+  }
+  return colors;
+}
+
 // Walk parents looking for a gradient background and return its color stops.
 // Used as a fallback when resolveBackground() returns null because the
 // effective background is a gradient (no single solid color to compare against).
+// Translucent solid layers found between the element and the gradient (frosted
+// panels, glass washes) are composited over every stop, the same way
+// resolveBackground flattens them over a solid base — raw stops alone would
+// false-flag dark text on a light frosted wash over a dark gradient, and miss
+// the inverse.
 function resolveGradientStops(el, win, customPropMap) {
   let current = el;
+  const overlays = [];
   while (current && current.nodeType === 1) {
     const style = DETECTOR_IS_BROWSER ? getComputedStyle(current) : win.getComputedStyle(current);
     const bgImage = style.backgroundImage || '';
+    // A url() layer anywhere in the stack — alone, or alongside a gradient in
+    // the same declaration (a translucent wash over a texture photo) — paints
+    // pixels the analytic walk cannot know. Measuring the gradient stops over
+    // the wrong base flagged dark ink sitting on a bright gold-leaf image at
+    // 2.6:1; skipping beats a wrong ratio, and the screenshot subsystem owns
+    // image-backed text.
+    if (bgImage && bgImage !== 'none' && /url\s*\(/i.test(bgImage)) return null;
     let stops = null;
     if (bgImage && bgImage !== 'none' && /gradient/i.test(bgImage)) {
-      const parsed = parseGradientColors(bgImage);
+      const parsed = parseGradientColorsModern(bgImage);
       if (parsed.length > 0) stops = parsed;
     }
     if (!stops && !DETECTOR_IS_BROWSER) {
-      // jsdom doesn't decompose `background:` shorthand — peek at the raw inline style
+      // Static mode: peek at the raw inline style for gradients the cascade did not surface
       const rawStyle = current.getAttribute?.('style') || '';
       const bgMatch = rawStyle.match(/background(?:-image)?\s*:\s*([^;]+)/i);
       if (bgMatch && /gradient/i.test(bgMatch[1])) {
-        const parsed = parseGradientColors(bgMatch[1]);
+        const parsed = parseGradientColorsModern(bgMatch[1]);
         if (parsed.length > 0) stops = parsed;
       }
     }
-    if (stops) return compositeGradientStops(stops, current, win, customPropMap);
+    if (stops) {
+      const composited = compositeGradientStops(stops, current, win, customPropMap);
+      if (!composited || overlays.length === 0) return composited;
+      return composited.map(stop => {
+        let acc = stop;
+        for (let i = overlays.length - 1; i >= 0; i--) acc = compositeColorOver(overlays[i], acc);
+        return acc;
+      });
+    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
+    if (bg && bg.a > 0.1) {
+      // An opaque surface above the gradient means the gradient never shows
+      // through here; resolveBackground would have returned it, so reaching
+      // this is defensive — bail rather than measure the wrong layer.
+      if (bg.a >= 0.99) return null;
+      overlays.push(bg);
+    }
     current = current.parentElement;
   }
   return null;
@@ -1986,6 +2109,10 @@ function checkElementColorsDOM(el) {
   const rect = el.getBoundingClientRect();
   if (rect.width < 10 || rect.height < 10) return [];
   const style = getComputedStyle(el);
+  // Invisible at rest: hidden scene variants (opacity-0 carousels, swap
+  // decks) are not user-visible, and measuring their inherited colors against
+  // whatever surface happens to sit behind the stack is noise, not audit.
+  if (style.visibility === 'hidden' || effectiveOpacityDOM(el) <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
   let effectiveBg = resolveBackground(el);
@@ -2798,11 +2925,13 @@ function checkElementGlowDOM(el) {
   // If resolveBackground returns null (gradient), try to infer from the gradient colors
   let parentBg = el.parentElement ? resolveBackground(el.parentElement) : resolveBackground(el);
   if (!parentBg) {
-    // Gradient background — sample its colors to determine if it's dark
+    // Gradient background — sample its colors to determine if it's dark.
+    // Modern-syntax parsing matters here: body-level gradients now reach this
+    // fallback in browser mode, and their stops usually serialize as oklch.
     let cur = el.parentElement;
     while (cur && cur.nodeType === 1) {
       const bgImage = getComputedStyle(cur).backgroundImage || '';
-      const gradColors = parseGradientColors(bgImage);
+      const gradColors = parseGradientColorsModern(bgImage);
       if (gradColors.length > 0) {
         // Average the gradient colors
         const avg = { r: 0, g: 0, b: 0 };
@@ -3644,6 +3773,14 @@ function checkElementBorders(tag, style, overrides, resolvedRadius, el = null) {
 }
 
 function checkElementColors(el, style, tag, window, customPropMap, hasAnchorInheritRule) {
+  // Invisible at rest, static twin of the browser walk's skip: opacity does
+  // not inherit, so walk ancestors multiplying declared opacity down.
+  if (style.visibility === 'hidden') return [];
+  let effOpacity = 1;
+  for (let cur = el; cur && cur.nodeType === 1 && effOpacity > 0.02; cur = cur.parentElement) {
+    effOpacity *= parseFloat(window.getComputedStyle(cur).opacity || '1');
+  }
+  if (effOpacity <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
 
@@ -4802,6 +4939,11 @@ function isRenderedForBrowserRule(el) {
 function checkElementTextOverflowDOM(el) {
   const tag = el.tagName.toLowerCase();
   if (TEXT_OVERFLOW_SKIP_TAGS.has(tag)) return [];
+  // scrollWidth/clientWidth are CSS box-model metrics; on SVG content Chrome
+  // returns arbitrary non-zero values for both (a <text> reported 78/48 while
+  // its rendered length sat comfortably inside its box), so the delta is
+  // noise, not overflow. SVG clips to its own viewport anyway.
+  if (el.namespaceURI === 'http://www.w3.org/2000/svg') return [];
   if (!isRenderedForBrowserRule(el)) return [];
   // Only the element that actually owns overflowing text — not its ancestors,
   // which inherit a wider scrollWidth from the spilling descendant.
@@ -5186,6 +5328,22 @@ function isPaintedForOcclusion(el) {
 // path is pure geometry and runs anywhere on the page.
 const OCCLUSION_TEXT_SKIP_TAGS = new Set(['script', 'style', 'noscript', 'template', 'title']);
 
+// An element whose effective opacity multiplies out to ~0 paints nothing at
+// rest: it is not user-visible, so visual findings on it (contrast, occlusion)
+// measure a state nobody sees. Browser-only — the walk needs live computed
+// styles. Cycling scenes that fade such elements in later are the screenshot
+// subsystem's territory, not the analytic walk's.
+function effectiveOpacityDOM(el) {
+  let o = 1;
+  // Walk all the way through body and html: `body { opacity: 0 }` page-fade
+  // wrappers hide every descendant just as thoroughly as a local wrapper.
+  for (let cur = el; cur && cur.nodeType === 1; cur = cur.parentElement) {
+    o *= parseFloat(getComputedStyle(cur).opacity || '1');
+    if (o <= 0.02) return 0;
+  }
+  return o;
+}
+
 function checkTextOcclusionDOM() {
   const findings = [];
   const seenVictims = new Set();
@@ -5213,6 +5371,11 @@ function checkTextOcclusionDOM() {
     }
     return false;
   };
+  // The classic occluder shape this rules out is an opacity-0 interaction
+  // layer — a range scrubber stretched over a before/after comparison — which
+  // elementFromPoint still returns and whose UA background-color otherwise
+  // reads as an opaque box.
+  const effectiveOpacity = effectiveOpacityDOM;
 
   // Collect renderable text owners in / near the first viewport for the
   // elementFromPoint probe. SVG <text> counts too.
@@ -5225,6 +5388,7 @@ function checkTextOcclusionDOM() {
     const text = inSvg ? (el.textContent || '').trim() : elementDirectText(el);
     if (text.length < 2) continue;
     if (!isPaintedForOcclusion(el)) continue;
+    if (effectiveOpacity(el) <= 0.02) continue;
     let rect; try { rect = el.getBoundingClientRect(); } catch { continue; }
     if (rect.width < 6 || rect.height < 6) continue;
     // Viewport-bound probe: keep text whose box overlaps the live viewport.
@@ -5258,6 +5422,7 @@ function checkTextOcclusionDOM() {
         if (top === el || el.contains(top) || top.contains(el)) continue;
         const topCs = getComputedStyle(top);
         if (isFloated(topCs) || isMarqueeish(top, topCs) || isPinnedOverlay(top)) continue;
+        if (effectiveOpacity(top) <= 0.02) continue;
         const topTag = top.tagName.toLowerCase();
         // Text sitting under a raw image/video is contrast territory (deduped
         // against the pixel low-contrast rule); leave those alone here.
@@ -5468,6 +5633,7 @@ export {
   CSS_NAMED_COLORS,
   checkBorders,
   isEmojiOnlyText,
+  scopedIgnoreActive,
   checkColors,
   checkHoverContrast,
   checkElementHoverContrast,

.agents/skills/impeccable/scripts/hook-before-edit.mjs

@@ -16,11 +16,15 @@ import path from 'node:path';
 
 import {
   ALLOWED_EXTS,
+  DEFAULT_CONFIG,
   EDIT_COUNT_THRESHOLD,
   GENERATED_PATH,
   SENSITIVE_PATH,
-  appendDesignSystemNote,
+  appendDesignSystemNoteOnce,
+  commitFooterShown,
+  designNoteReserve,
   designSystemOptions,
+  footerModeForSession,
   filterFindings,
   isNativePlatform,
   isScanTargetInsideProject,
@@ -345,13 +349,32 @@ async function detectProposedHtml(detector, content, filePath, scanOptions) {
   }
 }
 
-function cursorBlockMessage(findings, filePath, config, cwd) {
-  const rendered = renderTemplate(findings, filePath, config, { cwd });
-  const blocked = rendered.replace(
+// Cursor caps deny messages around 4000 chars. The cap feeds through the
+// renderer's clamp, which preserves the policy footer, rather than tail-
+// slicing the rendered text, which cut the footer off any message the
+// default 8000-char budget let past 4000.
+const CURSOR_DENY_LIMIT = 4000;
+const BLOCK_PREFIX = 'Impeccable design hook blocked this write before it landed. ';
+
+function cursorBlockMessage(findings, filePath, config, cwd, footerMode, reserveChars) {
+  const limits = config?.limits || DEFAULT_CONFIG.limits;
+  // Charge the prefix via reserveChars, not by subtracting from maxChars:
+  // renderTemplate's 500-char floor re-raises any maxChars pushed below it,
+  // un-charging a prefix subtracted from maxChars (Greptile P1 on PR #508).
+  // reserveChars comes off after the floor, so the prefix is charged at every
+  // config tier and the final prefixed message plus a pending staleness note
+  // fits the binding limit. Default-config output is byte-identical.
+  const budget = Math.min(
+    limits.maxChars || DEFAULT_CONFIG.limits.maxChars,
+    CURSOR_DENY_LIMIT,
+  );
+  const rendered = renderTemplate(findings, filePath,
+    { ...config, limits: { ...limits, maxChars: budget } },
+    { cwd, footer: footerMode, reserveChars: (reserveChars || 0) + BLOCK_PREFIX.length });
+  return rendered.replace(
     '[impeccable@1] Design hook findings requiring review',
-    '[impeccable@1] Impeccable design hook blocked this write before it landed. Design hook findings requiring review',
+    `[impeccable@1] ${BLOCK_PREFIX}Design hook findings requiring review`,
   );
-  return blocked.length > 4000 ? `${blocked.slice(0, 3984)}\n...(truncated)` : blocked;
 }
 
 function findingSignature(findings) {
@@ -468,9 +491,16 @@ async function main() {
     });
   }
 
-  const message = appendDesignSystemNote(cursorBlockMessage(filtered, filePath, config, cwd), scanOptions);
   const sessionId = event.session_id || event.conversation_id || 'unknown';
   const cache = readCache(cwd);
+  // Repeated denials for the same session repeat the findings, not the
+  // policy: the full footer emits once per session, the short form after.
+  const footerMode = footerModeForSession(cache, sessionId);
+  const message = appendDesignSystemNoteOnce(
+    cursorBlockMessage(filtered, filePath, config, cwd, footerMode, designNoteReserve(scanOptions, cache, sessionId)),
+    scanOptions, cache, sessionId, config,
+  );
+  commitFooterShown(cache, sessionId, message);
   const denial = bumpCursorDenial(cache, sessionId, filePath, filtered);
   persistCache(cwd, cache);
   if (denial.count > EDIT_COUNT_THRESHOLD) {

.agents/skills/impeccable/scripts/hook-lib.mjs

@@ -22,6 +22,9 @@
  *   dedupeAgainstCache(findings, cache, sessionId, filePath)
  *   renderTemplate(findings, filePath, config, opts)
  *   renderCleanAck(filePath, opts) / renderPendingAck(filePath, known, opts)
+ *   appendDesignSystemNote(text, scanOptions) / appendDesignSystemNoteOnce(text, scanOptions, cache, sessionId, config)
+ *   designNoteReserve(scanOptions, cache, sessionId)
+ *   footerModeForSession(cache, sessionId) / commitFooterShown(cache, sessionId, text)
  *   shouldEmitAckForFile(filePath, config?)
  *   writeAuditLog(env, entry)
  *   loadDetector() -> Promise<{ detectText, detectHtml }>
@@ -970,7 +973,13 @@ export function renderTemplate(findings, filePath, config, opts = {}) {
   if (!Array.isArray(findings) || findings.length === 0) return '';
   const limits = config?.limits || DEFAULT_CONFIG.limits;
   const cap = Math.max(1, limits.maxFindings || DEFAULT_CONFIG.limits.maxFindings);
-  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars);
+  // reserveChars holds back room for a note the caller appends after render
+  // (the DESIGN.md staleness note), so the final payload stays inside the
+  // configured budget. It comes off after the 500-char floor, so at floor
+  // configs the note keeps guaranteed delivery room; the clamp budget can
+  // therefore sit below 500, which clampLastLine's footer-preserving
+  // fallback handles (Bugbot on PR #508).
+  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars) - (opts.reserveChars || 0);
 
   const cwd = opts.cwd || process.cwd();
   const display = relativize(filePath, cwd);
@@ -979,11 +988,12 @@ export function renderTemplate(findings, filePath, config, opts = {}) {
   const remaining = total - shown.length;
 
   const header = `${ENVELOPE_PREFIX} Design hook findings requiring review in ${display} (${total} issue(s)):`;
-  const lines = shown.map((f) => formatFindingLine(f));
+  const seenRules = new Set();
+  const lines = shown.map((f) => formatDedupedFindingLine(f, seenRules));
   const more = remaining > 0
     ? `... and ${remaining} more (see ${IMPECCABLE_COMMAND} audit).`
     : null;
-  const footer = directiveFooter(display);
+  const footer = directiveFooter({ mode: opts.footer });
 
   const blocks = [header, ...lines];
   if (more) blocks.push(more);
@@ -1007,12 +1017,15 @@ function renderGroupedTemplate(groups, config, opts = {}) {
 
   const limits = config?.limits || DEFAULT_CONFIG.limits;
   const cap = Math.max(1, limits.maxFindings || DEFAULT_CONFIG.limits.maxFindings);
-  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars);
+  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars) - (opts.reserveChars || 0);
   const cwd = opts.cwd || process.cwd();
   const total = realGroups.reduce((sum, group) => sum + group.findings.length, 0);
   const header = `${ENVELOPE_PREFIX} Design hook findings requiring review across ${realGroups.length} files (${total} issue(s)):`;
   const lines = [];
   let shownCount = 0;
+  // One seen-set across all groups: a rule already described under one file
+  // is not re-described under the next.
+  const seenRules = new Set();
 
   for (const group of realGroups) {
     const display = relativize(group.filePath, cwd);
@@ -1020,7 +1033,7 @@ function renderGroupedTemplate(groups, config, opts = {}) {
     const remainingCap = Math.max(0, cap - shownCount);
     const shown = group.findings.slice(0, remainingCap);
     for (const finding of shown) {
-      lines.push(formatFindingLine(finding));
+      lines.push(formatDedupedFindingLine(finding, seenRules));
     }
     shownCount += shown.length;
     const hidden = group.findings.length - shown.length;
@@ -1029,7 +1042,7 @@ function renderGroupedTemplate(groups, config, opts = {}) {
     }
   }
 
-  const footer = directiveFooter('the affected files', { grouped: true });
+  const footer = directiveFooter({ mode: opts.footer });
   let text = [header, ...lines, '', footer].join('\n');
   if (text.length > maxChars) {
     text = clampGroupedToBudget(header, lines, footer, maxChars);
@@ -1037,82 +1050,149 @@ function renderGroupedTemplate(groups, config, opts = {}) {
   return text;
 }
 
+// The clamp contract, shared by both budget functions: the footer is policy,
+// not detail, so it survives every clamp. Try the requested footer first;
+// when it cannot fit even after dropping finding lines, retry with the short
+// policy rather than sacrifice findings that fit beside it. A result that
+// dropped every finding line (a grouped render can fit a bare file header)
+// does not count as a fit: findings are why the emission exists.
+const isFindingLine = (line) => line.startsWith('- ');
+
+function footerFallbacks(footer) {
+  const short = directiveFooter({ mode: 'short' });
+  return footer === short ? [footer] : [footer, short];
+}
+
 function clampGroupedToBudget(header, lines, footer, maxChars) {
-  const assemble = (linesArr, omitted) => [
+  const assemble = (linesArr, omitted, footerText) => [
     header,
     ...linesArr,
     ...(omitted ? [`... and more (see ${IMPECCABLE_COMMAND} audit).`] : []),
     '',
-    footer,
+    footerText,
   ].join('\n');
 
-  let working = lines.slice();
-  let omitted = false;
-  let assembled = assemble(working, omitted);
-  while (assembled.length > maxChars && working.length > 1) {
-    working.pop();
-    omitted = true;
-    assembled = assemble(working, omitted);
+  for (const footerText of footerFallbacks(footer)) {
+    let working = lines.slice();
+    let omitted = false;
+    let assembled = assemble(working, omitted, footerText);
+    while (assembled.length > maxChars && working.length > 1) {
+      working.pop();
+      omitted = true;
+      assembled = assemble(working, omitted, footerText);
+    }
+    if (assembled.length <= maxChars && working.some(isFindingLine)) return assembled;
   }
-  if (assembled.length > maxChars) {
-    assembled = `${assembled.slice(0, maxChars - 1)}…`;
-  }
-  return assembled;
+  return clampLastLine((linesArr, footerText) => assemble(linesArr, true, footerText),
+    lines.find(isFindingLine) || lines[0], maxChars);
 }
 
 function clampToBudget(header, lines, more, footer, maxChars) {
-  const assemble = (linesArr, moreText) => {
+  const assemble = (linesArr, moreText, footerText) => {
     const blocks = [header, ...linesArr];
     if (moreText) blocks.push(moreText);
     blocks.push('');
-    blocks.push(footer);
+    blocks.push(footerText);
     return blocks.join('\n');
   };
 
-  let working = lines.slice();
-  let moreText = more;
-  let assembled = assemble(working, moreText);
-  while (assembled.length > maxChars && working.length > 1) {
-    working.pop();
-    moreText = `... and more (see ${IMPECCABLE_COMMAND} audit).`;
-    assembled = assemble(working, moreText);
+  let lastMore = more;
+  for (const footerText of footerFallbacks(footer)) {
+    let working = lines.slice();
+    let moreText = more;
+    let assembled = assemble(working, moreText, footerText);
+    while (assembled.length > maxChars && working.length > 1) {
+      working.pop();
+      moreText = `... and more (see ${IMPECCABLE_COMMAND} audit).`;
+      assembled = assemble(working, moreText, footerText);
+    }
+    lastMore = moreText;
+    if (assembled.length <= maxChars) return assembled;
   }
-  if (assembled.length > maxChars) {
-    assembled = `${assembled.slice(0, maxChars - 1)}…`;
-  }
-  return assembled;
+  return clampLastLine((linesArr, footerText) => assemble(linesArr, lastMore, footerText),
+    lines.find(isFindingLine) || lines[0], maxChars);
 }
 
-function formatFindingLine(f) {
+// Last resort with one finding line left: the short policy gets the budget
+// first, the line is clipped to what remains. The pre-fix tail-slice cut
+// whatever happened to be last, which was always the footer.
+function clampLastLine(build, line, maxChars) {
+  const footerText = directiveFooter({ mode: 'short' });
+  const bare = build([], footerText);
+  // +1 for the newline the line itself brings when it joins the blocks.
+  const room = maxChars - bare.length - 1;
+  if (room >= 24) {
+    const clipped = line.length > room ? `${line.slice(0, room - 1)}…` : line;
+    return build([clipped], footerText);
+  }
+  // No room for even a clipped finding line: the note reservation can pull
+  // the budget below the 500-char floor, and a deep file path can push the
+  // header past what remains beside the short policy (Bugbot on PR #508).
+  // Drop the line, and if the bare header + policy still overflow, clip the
+  // head. Never tail-slice: the footer sits at the end, so a tail slice is
+  // exactly the footer cut this renderer exists to prevent.
+  if (bare.length <= maxChars) return bare;
+  const head = bare.slice(0, Math.max(0, maxChars - footerText.length - 4));
+  return `${head}…\n\n${footerText}`;
+}
+
+// `compact` drops the registry description: within one emission the first
+// occurrence of a rule carries the full description and repeats keep only the
+// rule id, name, and their own ignore hint (values differ per line, so the
+// hint must survive the dedupe).
+function formatFindingLine(f, opts = {}) {
   const prefix = f.line && f.line > 0 ? `- L${f.line}` : '-';
-  const desc = (f.description || '').trim();
+  const desc = opts.compact ? '' : (f.description || '').trim();
   const name = (f.name || '').trim();
   // Description from the registry already ends in punctuation; join with a
   // single space. `name` may have a trailing period already, keep it clean.
   const nameSegment = name ? `${name.replace(/\.+\s*$/, '')}.` : '';
-  const ignoreCommand = formatFindingIgnoreCommand(f);
-  const ignoreSegment = ignoreCommand
-    ? ` If the user explicitly confirms this value is intentional: \`${ignoreCommand}\`.`
-    : '';
+  const ignoreHint = formatFindingIgnoreHint(f);
+  const ignoreSegment = ignoreHint ? ` If intentional: \`${ignoreHint}\`.` : '';
   return `${prefix} [${f.antipattern}] ${nameSegment} ${desc}${ignoreSegment}`.replace(/\s+/g, ' ').trim();
 }
 
-function formatFindingIgnoreCommand(finding) {
+// Dedupe applied in shown-line order, so the first rendered occurrence of a
+// rule always carries the description. The budget clamps pop lines from the
+// end, which can never orphan a compact repeat before its described first
+// occurrence.
+function formatDedupedFindingLine(finding, seenRules) {
+  const rule = normalizeIgnoreRule(finding?.antipattern);
+  const compact = rule ? seenRules.has(rule) : false;
+  if (rule) seenRules.add(rule);
+  return formatFindingLine(finding, { compact });
+}
+
+// The rule/value pair the footer's `hook-admin.mjs ignore-value` command
+// takes. Deliberately just the args: the executable prefix, the --reason
+// contract, and the disclosure rule live in the directive footer, stated once
+// instead of per line.
+function formatFindingIgnoreHint(finding) {
   if (!finding || typeof finding !== 'object') return '';
   const rule = normalizeIgnoreRule(finding.antipattern);
   if (!rule) return '';
   const normalizedValue = extractFindingIgnoreValue(finding);
   if (!normalizedValue) return '';
-  const value = extractFindingIgnoreValueRaw(finding);
-  const valueArg = quoteCommandArg(value);
-  const reason = quoteCommandArg(`User confirmed ${value} is intentional`);
-  return `${IMPECCABLE_COMMAND} hooks ignore-value ${rule} ${valueArg} --shared --reason ${reason}`;
+  const valueArg = quoteCommandArg(extractFindingIgnoreValueRaw(finding));
+  return `ignore-value ${rule} ${valueArg}`;
 }
 
 function quoteCommandArg(value) {
   const text = String(value || '').trim();
   if (/^[A-Za-z0-9._:-]+$/.test(text)) return text;
-  return `"${text.replace(/\\/g, '\\\\').replace(/"/g, '\\"')}"`;
+  // The suggestion is meant to be run on this same machine, so quote for its
+  // shell. POSIX /bin/sh still expands $(...), backticks, and ${} inside
+  // double quotes, and these values come from scanned file content (a
+  // font-family name) or a file path, so untrusted input must be
+  // single-quoted (issue #476). Windows cmd.exe performs no such command
+  // substitution, but it treats a single quote as a literal character rather
+  // than a grouping delimiter, so a value or path containing spaces has to
+  // stay double-quoted there (Greptile #533). Keep the pre-existing
+  // double-quote escaping on Windows so that path's behavior is unchanged.
+  if (process.platform === 'win32') {
+    return `"${text.replace(/\\/g, '\\\\').replace(/"/g, '\\"')}"`;
+  }
+  return `'${text.replace(/'/g, `'\\''`)}'`;
 }
 
 function relativize(filePath, cwd) {
@@ -1594,36 +1674,105 @@ export function designSystemOptions(config, detector, projectCwd) {
   }
 }
 
+const DESIGN_STALE_NOTE = `${ENVELOPE_PREFIX} DESIGN.md is newer than .impeccable/design.json. Run ${IMPECCABLE_COMMAND} document to refresh the design-system sidecar.`;
+
 export function appendDesignSystemNote(text, scanOptions) {
   if (!text || !scanOptions?.designSystem?.mdNewerThanJson) return text;
-  return `${text}\n\n${ENVELOPE_PREFIX} DESIGN.md is newer than .impeccable/design.json. Run ${IMPECCABLE_COMMAND} document to refresh the design-system sidecar.`;
+  return `${text}\n\n${DESIGN_STALE_NOTE}`;
 }
 
+// Session-scoped once-only gate for repeat-prone message parts. Returns true
+// the first time a flag is consumed in a session and false after, mirroring
+// the `cleanAcked` mechanic: the mtime skew (and the policy footer) do not
+// change between edits, so re-stating them on every emission spends context
+// to say nothing new. Callers must persist the cache for the flag to stick.
+function consumeSessionNoticeFlag(cache, sessionId, flag) {
+  const session = ensureSession(cache, sessionId);
+  if (session[flag]) return false;
+  session[flag] = true;
+  session.updatedAt = Date.now();
+  return true;
+}
+
+// Once-per-session variant of appendDesignSystemNote for the emission paths
+// that have cache access. The staleness note names standing project state,
+// not new information, so one mention per session is enough. The note is
+// appended after the renderer has clamped to the configured budget: render
+// paths reserve room for it via designNoteReserve, and the size check here
+// is the safety net for the ack paths, deferring (without consuming the
+// flag) to a later emission rather than busting maxChars.
+export function appendDesignSystemNoteOnce(text, scanOptions, cache, sessionId, config) {
+  if (!text || !scanOptions?.designSystem?.mdNewerThanJson) return text;
+  const maxChars = Math.max(500, config?.limits?.maxChars || DEFAULT_CONFIG.limits.maxChars);
+  if (text.length + DESIGN_STALE_NOTE.length + 2 > maxChars) return text;
+  if (!consumeSessionNoticeFlag(cache, sessionId, 'designNoteShown')) return text;
+  return appendDesignSystemNote(text, scanOptions);
+}
+
+// Render-time reservation for the note above: how many characters the
+// renderer must hold back so a pending staleness note still fits inside the
+// configured budget. Zero once the session has seen the note. Without the
+// reservation, a session whose every emission fills the budget would defer
+// the note forever.
+export function designNoteReserve(scanOptions, cache, sessionId) {
+  if (!scanOptions?.designSystem?.mdNewerThanJson) return 0;
+  if (ensureSession(cache, sessionId).designNoteShown) return 0;
+  return DESIGN_STALE_NOTE.length + 2;
+}
+
+// Full directive footer once per session, the short reminder after. Fresh
+// emissions and Cursor denials share the session flag (`footerShown`), so a
+// session pays the full policy exactly once however it first fires. The mode
+// is a peek: the clamp can downgrade a requested full footer under a tight
+// budget, so the flag commits only when the complete full policy actually
+// reached the output. Matching the whole footer text (not a sentinel) keeps
+// the flag honest against any truncation that spares the opening words.
+export function footerModeForSession(cache, sessionId) {
+  return ensureSession(cache, sessionId).footerShown ? 'short' : 'full';
+}
+
+export function commitFooterShown(cache, sessionId, text) {
+  if (!text || !text.includes(directiveFooter())) return;
+  const session = ensureSession(cache, sessionId);
+  if (session.footerShown) return;
+  session.footerShown = true;
+  session.updatedAt = Date.now();
+}
+
+const HOOK_ADMIN_COMMAND = `node ${quoteCommandArg(path.join(__dirname, 'hook-admin.mjs'))}`;
+
 // The directive footer is the part of the hook output that steers model
-// behavior. Three intentional moves:
-//   1. **Imperative, not advisory.** "Handle these..." beats "Consider
-//      revising..." which the model treats as a soft suggestion it can
-//      override when the user asked for any kind of throwaway / demo UI.
-//   2. **Explicit judgment clause.** Without it, the model will try to
-//      "fix" intentional motion, bad fixtures, anti-pattern examples in
-//      docs, or test cases. Naming the judgment inline beats hoping the
-//      model infers it from context.
-//   3. **Acknowledgement instruction.** Hook output is injected as
-//      developer-role context, not a chat turn, so the user never sees the
-//      raw envelope. Asking the model to surface the resolution in its
-//      reply is the cheapest way to make the feedback loop visible.
-function directiveFooter(display, opts = {}) {
-  // Offer the rule-scoped-to-file form first. `ignore-file` silences every rule
-  // for the path forever, which is far more than one noisy rule on a real UI
-  // surface justifies, and it was previously the only option named here.
-  const target = opts.grouped ? '<path>' : quoteCommandArg(display);
-  const fileIgnoreGuidance = `run \`${IMPECCABLE_COMMAND} hooks ignore-value <id> "*" --file ${target}\` to scope just that rule to the file, or \`${IMPECCABLE_COMMAND} hooks ignore-file ${target}\` only when the whole file is out of scope for design review (a fixture, a generated artifact, a deliberate demo)`;
+// behavior. Intentional moves, in order:
+//   1. **Imperative, not advisory.** "Triage each finding..." beats
+//      "Consider revising...", which the model treats as a soft suggestion.
+//   2. **Positive triage branches.** Fix / suppress-and-disclose / ask. The
+//      suppress branch names the calibration examples (demo, fixture,
+//      documented bad design, user-confirmed choice) because the agent now
+//      acts on its own confidence and needs the bar stated.
+//   3. **Executable ignore path.** The old footer named only the slash
+//      command, which an agent reacting to hook output cannot run; the
+//      hook-admin.mjs invocation is runnable as-is and keeps agents out of
+//      hand-editing config.json.
+//   4. **Honest provenance.** The --reason is the audit trail; "user
+//      confirmed" appears only when the user actually did.
+//   5. **Acknowledgement instruction.** Hook output is injected as
+//      developer-role context, so the reply is where the user sees the
+//      resolution, including any ignore the agent persisted.
+//   6. **Once per session.** The full policy emits on the session's first
+//      fire; later emissions carry the one-line short form (mode 'short').
+function directiveFooter(opts = {}) {
+  if (opts.mode === 'short') {
+    // No command path here: the session's first emission already gave the
+    // runnable hook-admin.mjs invocation, and restating ~70 chars of absolute
+    // path on every repeat is the duplication this mode exists to cut.
+    return 'Triage per the session policy: fix real problems; persist confident false-positive or sanctioned-exception ignores via `hook-admin.mjs ignore-value` and disclose them in your reply; unsure, ask in one line.';
+  }
   return [
-    'Handle these before finalizing: fix findings that are real design problems, or explicitly classify contextually intentional findings as false positives. Acknowledge what you changed or why you are leaving a finding unchanged.',
-    '',
-    'Use context judgment before editing. A finding is not automatically a defect; literal or domain-appropriate motion, intentional demos or fixtures, documentation of bad design, and user-confirmed choices can be valid as-is.',
-    '',
-    `Do not change intentional design just to satisfy the hook, and do not silence a real finding with an inline ignore comment to skip fixing it. Suppress a finding only after the user explicitly confirms it is intentional. Prefer a config ignore (one reviewable place, the commands below); reach for an inline \`impeccable-disable <rule>\` comment only when the waiver must travel with a file that leaves the repo, such as an exported or standalone document. Prefer the narrowest persisted exception: run the exact \`${IMPECCABLE_COMMAND} hooks ignore-value ... --shared\` command shown next to a value-specific finding. For \`overused-font\`, use \`ignore-value\` for a specific font and use \`${IMPECCABLE_COMMAND} hooks ignore-rule overused-font --all-values\` only when the user asks to ignore overused fonts generally. For a finding whose line shows no exact ignore-value command, such as \`side-tab\`, ${fileIgnoreGuidance}; use \`${IMPECCABLE_COMMAND} hooks ignore-rule <id>\` only when the user asks to suppress the whole non-value-specific rule. Run ${IMPECCABLE_COMMAND} audit for the full pass.`,
+    'Triage each finding, then state in your reply what you fixed, what you suppressed, and what you left standing:',
+    '- Real design problem: fix it. Keep intentional design as designed.',
+    `- Confident false positive or sanctioned exception (an intentional demo or fixture, documentation of bad design, literal or domain-appropriate motion, a choice the user confirmed): persist the narrowest ignore yourself and disclose it. Run \`${HOOK_ADMIN_COMMAND} ignore-value <rule> "<value>" --reason "<who decided: evidence>"\` with the pair shown on the finding line, or value "*" plus \`--file <path>\` when the line shows none. Write "user confirmed" in a reason only when the user did.`,
+    '- Unsure: leave it as is and ask the user in one line.',
+    `Self-serve ends at ignore-value: \`ignore-file\` and \`ignore-rule\` need the user's explicit approval, and never add an ignore to push a blocked write through. Full suppression ladder: ${IMPECCABLE_COMMAND} hooks.`,
   ].join('\n');
 }
 
@@ -1845,20 +1994,23 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       }
     }
 
-    // Persist only when the write is earned: fresh findings justify creating
-    // `.impeccable/` (dedup and suppression need it), deferred findings do
-    // too (the Stop deep pass needs the touched-file list to surface them),
-    // and an already-present `.impeccable/` dir marks a project that opted
-    // in. A non-UI edit, or a clean UI edit in a project with no Impeccable
-    // footprint, must be a no-op on disk (issues #344, #305).
-    if (freshGroups.length > 0 || deferredTotal > 0
-      || (cacheDirty && fs.existsSync(path.join(projectCwd, '.impeccable')))) {
-      persistCache(projectCwd, cache);
-    }
-
+    // The session notice flags mutate the cache, so they must settle before
+    // the persist that makes them stick across events.
     if (freshGroups.length > 0) {
       const firstGroup = freshGroups[0];
-      const text = appendDesignSystemNote(renderGroupedTemplate(freshGroups, config, { cwd: projectCwd }), scanOptions);
+      const footerMode = footerModeForSession(cache, sessionId);
+      const text = appendDesignSystemNoteOnce(
+        renderGroupedTemplate(freshGroups, config, {
+          cwd: projectCwd,
+          footer: footerMode,
+          reserveChars: designNoteReserve(scanOptions, cache, sessionId),
+        }),
+        scanOptions, cache, sessionId, config,
+      );
+      commitFooterShown(cache, sessionId, text);
+      // Fresh findings always earn the cache write, including creating
+      // `.impeccable/`: dedup, suppression, and the notice flags need it.
+      persistCache(projectCwd, cache);
       const allFindings = freshGroups.flatMap((group) => group.findings);
       return {
         exitCode: 0,
@@ -1881,6 +2033,33 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       };
     }
 
+    // Resolve the ack emission before the persist below: appendDesignSystem-
+    // NoteOnce consumes a session flag, and the flag only sticks when the
+    // write happens after it. Quiet mode emits nothing, so it consumes
+    // nothing. The clean arm mirrors the branch order further down: pending
+    // outranks suppression, suppression outranks clean.
+    let ack = null;
+    if (!quietMode && pendingWinner && shouldEmitAckForFile(pendingWinner.filePath, config)) {
+      ack = {
+        kind: 'pending',
+        text: appendDesignSystemNoteOnce(renderPendingAck(pendingWinner.filePath, pendingWinner.known, { cwd: projectCwd }), scanOptions, cache, sessionId, config),
+      };
+    } else if (!quietMode && !suppressionWinner && cleanWinner && !cleanAckDeduped && shouldEmitAckForFile(cleanWinner.filePath, config)) {
+      ack = {
+        kind: 'clean',
+        text: appendDesignSystemNoteOnce(renderCleanAck(cleanWinner.filePath, { cwd: projectCwd }), scanOptions, cache, sessionId, config),
+      };
+    }
+
+    // Persist only when the write is earned: deferred findings need the
+    // touched-file list for the Stop deep pass, and an already-present
+    // `.impeccable/` dir marks a project that opted in. A non-UI edit, or a
+    // clean UI edit in a project with no Impeccable footprint, must be a
+    // no-op on disk (issues #344, #305).
+    if (deferredTotal > 0 || (cacheDirty && fs.existsSync(path.join(projectCwd, '.impeccable')))) {
+      persistCache(projectCwd, cache);
+    }
+
     if (detectorThrewAny && !pendingWinner && !cleanWinner) {
       return result({ emitted: false, error: 'detector-threw', durationMs: Date.now() - started });
     }
@@ -1889,8 +2068,8 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       return result({ emitted: false, quiet: true, durationMs: Date.now() - started });
     }
 
-    if (pendingWinner && shouldEmitAckForFile(pendingWinner.filePath, config)) {
-      const text = appendDesignSystemNote(renderPendingAck(pendingWinner.filePath, pendingWinner.known, { cwd: projectCwd }), scanOptions);
+    if (ack?.kind === 'pending') {
+      const text = ack.text;
       return {
         exitCode: 0,
         stdout: payload(text, 'PostToolUse', harness),
@@ -1923,8 +2102,8 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       };
     }
 
-    if (cleanWinner && !cleanAckDeduped && shouldEmitAckForFile(cleanWinner.filePath, config)) {
-      const text = appendDesignSystemNote(renderCleanAck(cleanWinner.filePath, { cwd: projectCwd }), scanOptions);
+    if (ack?.kind === 'clean') {
+      const text = ack.text;
       return {
         exitCode: 0,
         stdout: payload(text, 'PostToolUse', harness),
@@ -2108,11 +2287,22 @@ export async function runStopHook({ stdinJson, env = {}, cwd = process.cwd(), no
       return result({ emitted: false, skipped: 'stop-clean', durationMs: Date.now() - started });
     }
 
-    // Fresh findings earn the cache write so the next Stop fire is silent
-    // unless new issues appear.
-    persistCache(projectCwd, cache);
+    // A per-edit fire earlier in this session already consumed the footer
+    // flag, so the Stop wall of text carries the one-line short footer.
+    const footerMode = footerModeForSession(cache, sessionId);
+    const text = appendDesignSystemNoteOnce(
+      renderGroupedTemplate(freshGroups, config, {
+        cwd: projectCwd,
+        footer: footerMode,
+        reserveChars: designNoteReserve(scanOptions, cache, sessionId),
+      }),
+      scanOptions, cache, sessionId, config,
+    );
+    commitFooterShown(cache, sessionId, text);
 
-    const text = appendDesignSystemNote(renderGroupedTemplate(freshGroups, config, { cwd: projectCwd }), scanOptions);
+    // Fresh findings earn the cache write so the next Stop fire is silent
+    // unless new issues appear; the notice flags ride along.
+    persistCache(projectCwd, cache);
     return {
       exitCode: 0,
       stdout: payload(text, 'Stop', harness),

.agents/skills/impeccable/scripts/lib/concept-catalog.mjs

@@ -109,6 +109,18 @@ export function validateConceptEntry(concept, { existingForms = new Map(), axes
     || concept.tags.some(tag => typeof tag !== 'string' || !tag.trim())) {
     errors.push(`concept ${id} must have exactly three structural tags`);
   }
+  // The slop this world in particular is at risk of. Optional, because 541
+  // entries predate it and none of them are wrong for lacking it. A world built
+  // from posters is at risk of shouting and one built from instruments is at
+  // risk of dead greys; a global detector cannot know which, and the author can.
+  if (concept?.avoid !== undefined) {
+    if (!Array.isArray(concept.avoid)
+      || concept.avoid.length < 2
+      || concept.avoid.length > 3
+      || concept.avoid.some(item => typeof item !== 'string' || item.trim().length < 12 || item.trim().length > 160)) {
+      errors.push(`concept ${id} avoid must be two or three negations of 12–160 characters`);
+    }
+  }
   if (!Array.isArray(concept?.system)
     || concept.system.length !== SYSTEM_PREFIXES.length
     || concept.system.some(rule => typeof rule !== 'string' || rule.trim().length < 12 || rule.trim().length > 180)) {

.agents/skills/impeccable/scripts/lib/impeccable-config.mjs

@@ -206,10 +206,10 @@ function parseIgnoreColor(value) {
   if (rgb) {
     const parts = splitColorArgs(rgb[1]);
     if (parts.length < 3 || parts.length > 4) return null;
-    const r = parseRgbChannel(parts[0]);
-    const g = parseRgbChannel(parts[1]);
-    const b = parseRgbChannel(parts[2]);
-    const a = parts[3] === undefined ? 1 : parseAlphaChannel(parts[3]);
+    const r = parseColorChannel(parts[0], COLOR_CHANNEL_FORMATS.rgb);
+    const g = parseColorChannel(parts[1], COLOR_CHANNEL_FORMATS.rgb);
+    const b = parseColorChannel(parts[2], COLOR_CHANNEL_FORMATS.rgb);
+    const a = parts[3] === undefined ? 1 : parseColorChannel(parts[3], COLOR_CHANNEL_FORMATS.alpha);
     if ([r, g, b, a].some((v) => v === null)) return null;
     return { r, g, b, a };
   }
@@ -218,10 +218,10 @@ function parseIgnoreColor(value) {
   if (hsl) {
     const parts = splitColorArgs(hsl[1]);
     if (parts.length < 3 || parts.length > 4) return null;
-    const h = parseHueChannel(parts[0]);
-    const s = parsePercentChannel(parts[1]);
-    const l = parsePercentChannel(parts[2]);
-    const a = parts[3] === undefined ? 1 : parseAlphaChannel(parts[3]);
+    const h = parseColorChannel(parts[0], COLOR_CHANNEL_FORMATS.hue);
+    const s = parseColorChannel(parts[1], COLOR_CHANNEL_FORMATS.percent);
+    const l = parseColorChannel(parts[2], COLOR_CHANNEL_FORMATS.percent);
+    const a = parts[3] === undefined ? 1 : parseColorChannel(parts[3], COLOR_CHANNEL_FORMATS.alpha);
     if ([h, s, l, a].some((v) => v === null)) return null;
     return hslToRgb(h, s, l, a);
   }
@@ -230,18 +230,13 @@ function parseIgnoreColor(value) {
 }
 
 function parseHexIgnoreColor(hex) {
-  if (hex.length === 3 || hex.length === 4) {
-    const r = parseInt(hex[0] + hex[0], 16);
-    const g = parseInt(hex[1] + hex[1], 16);
-    const b = parseInt(hex[2] + hex[2], 16);
-    const a = hex.length === 4 ? parseInt(hex[3] + hex[3], 16) / 255 : 1;
-    return { r, g, b, a };
-  }
-  const r = parseInt(hex.slice(0, 2), 16);
-  const g = parseInt(hex.slice(2, 4), 16);
-  const b = parseInt(hex.slice(4, 6), 16);
-  const a = hex.length === 8 ? parseInt(hex.slice(6, 8), 16) / 255 : 1;
-  return { r, g, b, a };
+  const expanded = hex.length <= 4
+    ? [...hex].map((digit) => digit.repeat(2)).join('')
+    : hex;
+  const [r, g, b, alpha = 255] = expanded
+    .match(/../g)
+    .map((channel) => Number.parseInt(channel, 16));
+  return { r, g, b, a: alpha / 255 };
 }
 
 function splitColorArgs(body) {
@@ -259,47 +254,34 @@ function splitColorArgs(body) {
   return text.replace(/\s*\/\s*/g, ' / ').split(/\s+/).filter((part) => part && part !== '/');
 }
 
-function parseRgbChannel(raw) {
-  const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)(%)?$/);
-  if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  const scaled = match[2] ? value * 2.55 : value;
-  if (scaled < 0 || scaled > 255) return null;
-  return Math.round(scaled);
-}
+const CSS_NUMBER_RE = /^(-?\d*\.?\d+)(%|deg|rad|turn|grad)?$/;
+const identity = (value) => value;
+const COLOR_CHANNEL_FORMATS = {
+  rgb: { units: { '': identity, '%': (value) => value * 2.55 }, min: 0, max: 255, round: true },
+  alpha: { units: { '': identity, '%': (value) => value / 100 }, min: 0, max: 1 },
+  hue: {
+    units: {
+      '': identity,
+      deg: identity,
+      rad: (value) => value * (180 / Math.PI),
+      turn: (value) => value * 360,
+      grad: (value) => value * 0.9,
+    },
+  },
+  percent: { units: { '%': (value) => value / 100 }, min: 0, max: 1 },
+};
 
-function parseAlphaChannel(raw) {
+function parseColorChannel(raw, { units, min = -Infinity, max = Infinity, round = false }) {
   const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)(%)?$/);
+  const match = text.match(CSS_NUMBER_RE);
   if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  const alpha = match[2] ? value / 100 : value;
-  return alpha >= 0 && alpha <= 1 ? alpha : null;
-}
-
-function parseHueChannel(raw) {
-  const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)(deg|rad|turn|grad)?$/);
-  if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  const unit = match[2] || 'deg';
-  if (unit === 'turn') return value * 360;
-  if (unit === 'rad') return value * (180 / Math.PI);
-  if (unit === 'grad') return value * 0.9;
-  return value;
-}
-
-function parsePercentChannel(raw) {
-  const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)%$/);
-  if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  return value >= 0 && value <= 100 ? value / 100 : null;
+  const convert = units[match[2] || ''];
+  if (!convert) return null;
+  const number = Number.parseFloat(match[1]);
+  if (!Number.isFinite(number)) return null;
+  const value = convert(number);
+  if (value < min || value > max) return null;
+  return round ? Math.round(value) : value;
 }
 
 function hslToRgb(hue, saturation, lightness, alpha) {

.agents/skills/impeccable/scripts/lib/is-generated.mjs

@@ -13,7 +13,7 @@
  *      within the first ~300 characters — catches non-git projects.
  */
 
-import { execSync } from 'node:child_process';
+import { execFileSync } from 'node:child_process';
 import fs from 'node:fs';
 import path from 'node:path';
 
@@ -41,7 +41,10 @@ export function isGeneratedFile(filePath, options = {}) {
 
 function isGitIgnored(absPath, cwd) {
   try {
-    execSync(`git check-ignore --quiet ${JSON.stringify(absPath)}`, {
+    // argv form, never a shell: this runs on every file the live-mode source
+    // walk reaches, so a hostile filename embedding $(...) or backticks must
+    // not be interpretable (issue #476). JSON.stringify is not shell quoting.
+    execFileSync('git', ['check-ignore', '--quiet', absPath], {
       cwd,
       stdio: 'ignore',
     });

.agents/skills/impeccable/scripts/lib/open-system-browser.mjs

@@ -0,0 +1,26 @@
+import { spawn } from 'node:child_process';
+
+export function browserOpenCommand(url, {
+  platform = process.platform,
+  comspec = process.env.ComSpec || process.env.COMSPEC || 'cmd.exe',
+} = {}) {
+  if (platform === 'darwin') return { command: 'open', args: [url] };
+  if (platform === 'win32') return { command: comspec, args: ['/c', 'start', '', url] };
+  return { command: 'xdg-open', args: [url] };
+}
+
+export function openSystemBrowser(url, {
+  platform = process.platform,
+  comspec = process.env.ComSpec || process.env.COMSPEC || 'cmd.exe',
+  spawnImpl = spawn,
+} = {}) {
+  const { command, args } = browserOpenCommand(url, { platform, comspec });
+  try {
+    const child = spawnImpl(command, args, { stdio: 'ignore', detached: true });
+    child.on('error', () => {});
+    child.unref();
+    return true;
+  } catch {
+    return false;
+  }
+}

.agents/skills/impeccable/scripts/lib/staleness-deep.mjs

@@ -244,7 +244,8 @@ const HOOK_MARKER = /skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs/;
 //   * bundle-relative: node ".agents/.../hook.mjs"
 //   * legacy unquoted: node .claude/.../hook.mjs
 //   * guarded (#399):  [ ! -f "PATH" ] || node "PATH"   (PATH twice, identical)
-//   * absolute:        node "/Users/.../hook.mjs"        (user-level installs)
+//   * absolute (#476): [ ! -f 'PATH' ] || node 'PATH'   (single-quoted since
+//                      the shell-injection fix; older installs double-quote)
 //   * github portable: node "$(git rev-parse --show-toplevel)/.../hook.mjs"
 // A quoted path wins; the guard's two occurrences are identical, so the first
 // quoted match is the path. Otherwise fall back to the whitespace/metachar-
@@ -255,6 +256,12 @@ function hookScriptTokenFrom(command) {
   if (!HOOK_MARKER.test(str)) return null;
   const quoted = str.match(/"([^"]*skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs)"/);
   if (quoted) return quoted[1];
+  // A path containing an apostrophe serializes as '\'' inside single quotes;
+  // no regex reassembles that, and the bare fallback would misread a fragment
+  // of it, so return null: the caller never asserts on a path it can't parse.
+  if (str.includes("'\\''")) return null;
+  const singleQuoted = str.match(/'([^']*skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs)'/);
+  if (singleQuoted) return singleQuoted[1];
   const bare = str.match(/([^\s"'|&;()]*skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs)/);
   return bare ? bare[1] : null;
 }

.agents/skills/impeccable/scripts/live-browser.js

@@ -97,23 +97,20 @@
     return { value: c.value, label: c.label };
   });
 
-  const LIVE_CHROME_MOUNT_CONTRACT = ['root', 'transport', 'state', 'actions'];
-  const LIVE_UI_SURFACES = [
-    { key: 'global-bottom-bar', ids: [PREFIX + '-global-bar', PREFIX + '-global-bar-brand', PREFIX + '-pick-toggle', PREFIX + '-insert-toggle', PREFIX + '-detect-toggle', PREFIX + '-detect-badge', PREFIX + '-design-toggle', PREFIX + '-page-chat', PREFIX + '-page-chat-input', PREFIX + '-page-chat-voice', PREFIX + '-page-chat-send'] },
-    { key: 'pending-copy-edit-dock', ids: [PREFIX + '-pending-dock'] },
-    { key: 'element-selection-chrome', ids: [PREFIX + '-highlight', PREFIX + '-tooltip', PREFIX + '-bar', PREFIX + '-selection-pill', PREFIX + '-input', PREFIX + '-configure-voice', PREFIX + '-configure-bar-tooltip'] },
-    { key: 'action-picker', ids: [PREFIX + '-picker'] },
-    { key: 'edit-chrome', ids: [PREFIX + '-edit-badge'] },
-    { key: 'generating-row', ids: [PREFIX + '-bar', PREFIX + '-shader'] },
-    { key: 'variant-cycling-row', ids: [PREFIX + '-bar', PREFIX + '-params-panel'] },
-    { key: 'variant-params-panel', ids: [PREFIX + '-params-panel'] },
-    { key: 'saving-confirmed-rows', ids: [PREFIX + '-bar'] },
-    { key: 'insert-mode-chrome', ids: [PREFIX + '-insert-line', PREFIX + '-insert-placeholder', PREFIX + '-placeholder-resize', PREFIX + '-insert-input', PREFIX + '-insert-voice', PREFIX + '-insert-create', PREFIX + '-insert-create-tooltip'] },
-    { key: 'annotation-chrome', ids: [PREFIX + '-annot', PREFIX + '-annot-svg', PREFIX + '-annot-pins', PREFIX + '-annot-clear'] },
-    { key: 'design-system-panel', ids: [PREFIX + '-design-host'] },
-    { key: 'toasts-and-errors', ids: [PREFIX + '-toast', PREFIX + '-mount-error'] },
-    { key: 'css-isolation-boundary', ids: [PREFIX + '-root'] },
-  ];
+  // The Live chrome inventory (which surfaces exist, and the element ids each
+  // one owns) comes from the canonical source, skill/scripts/live/ui-surfaces.mjs,
+  // which the /live.js assembler serializes into these globals alongside the
+  // token/port/vocabulary. This file is served raw and injected as a classic
+  // script, so it cannot import that module; the private impeccable-site repo
+  // imports it directly to check its Live UI lab holds a snapshot for every
+  // surface, which only works while the list has exactly one definition.
+  // Add a surface in ui-surfaces.mjs, not here.
+  const LIVE_CHROME_MOUNT_CONTRACT = Array.isArray(window.__IMPECCABLE_LIVE_MOUNT_CONTRACT__)
+    ? window.__IMPECCABLE_LIVE_MOUNT_CONTRACT__
+    : ['root', 'transport', 'state', 'actions'];
+  const LIVE_UI_SURFACES = Array.isArray(window.__IMPECCABLE_LIVE_UI_SURFACES__)
+    ? window.__IMPECCABLE_LIVE_UI_SURFACES__
+    : [];
   const LIVE_UI_COMPONENT_IDS = [...new Set(LIVE_UI_SURFACES.flatMap((surface) => surface.ids))];
 
   //

.agents/skills/impeccable/scripts/live-copy-edit-agent.mjs

@@ -14,10 +14,12 @@ import path from 'node:path';
 import { createRequire } from 'node:module';
 
 const DEFAULT_TIMEOUT_MS = 60_000;
+const BATCH_OP_TEXT_LIMIT = 240;
 const require = createRequire(import.meta.url);
 
 export function buildCopyEditBatchPrompt(batch, { cwd = process.cwd() } = {}) {
-  const repairLines = batch?.repair ? [
+  const compactBatch = compactBatchForPrompt(batch);
+  const repairLines = compactBatch.repair ? [
     '',
     'Repair mode:',
     '- The previous Apply attempt changed source, but validation failed.',
@@ -28,7 +30,7 @@ export function buildCopyEditBatchPrompt(batch, { cwd = process.cwd() } = {}) {
     '- If failures or candidates show edited text is also a lookup key, update coupled count, animation, icon, image, asset, style, or metadata keys in the current source, or fail that entry without partial edits.',
     '- Keep failed and notes as arrays.',
     '- Return the same canonical JSON shape after repair.',
-    JSON.stringify(batch.repair, null, 2),
+    JSON.stringify(compactBatch.repair, null, 2),
   ] : [];
   return [
     'You are the Impeccable staged copy-edit batch applier.',
@@ -80,7 +82,7 @@ export function buildCopyEditBatchPrompt(batch, { cwd = process.cwd() } = {}) {
     ...repairLines,
     '',
     'Staged copy-edit batch:',
-    JSON.stringify(compactBatchForPrompt(batch), null, 2),
+    JSON.stringify(compactBatch, null, 2),
   ].join('\n');
 }
 
@@ -292,7 +294,7 @@ function readManualEditValidationScript(cwd) {
 function compactBatchForPrompt(batch) {
   return {
     pageUrl: batch?.pageUrl || null,
-    repair: batch?.repair || undefined,
+    repair: compactBatchRepair(batch?.repair),
     entries: (batch?.entries || []).map((entry) => ({
       id: entry.id,
       pageUrl: entry.pageUrl,
@@ -300,7 +302,71 @@ function compactBatchForPrompt(batch) {
       element: compactContextForBatch(entry.element),
       ops: (entry.ops || []).map(compactBatchOp),
     })),
-    candidates: batch?.candidates || [],
+    candidates: compactBatchCandidates(batch?.candidates),
+  };
+}
+
+function compactBatchRepair(repair) {
+  if (!repair || typeof repair !== 'object') return undefined;
+  return {
+    status: compactBatchString(repair.status),
+    attempt: normalizeOptionalBatchNumber(repair.attempt),
+    attempts: normalizeOptionalBatchNumber(repair.attempts),
+    maxAttempts: normalizeOptionalBatchNumber(repair.maxAttempts),
+    reason: compactBatchString(repair.reason),
+    transactionId: compactBatchString(repair.transactionId),
+    pageUrl: compactBatchString(repair.pageUrl),
+    failures: compactBatchDiagnostics(repair.failures),
+    files: compactBatchStringList(repair.files, 20),
+  };
+}
+
+function compactBatchDiagnostics(items, depth = 0) {
+  if (!Array.isArray(items)) return undefined;
+  return items.slice(0, 12).map((item) => ({
+    entryId: compactBatchString(item?.entryId || item?.id),
+    reason: compactBatchString(item?.reason || item?.kind),
+    detail: compactBatchString(item?.detail),
+    message: compactBatchString(item?.message),
+    file: compactBatchString(item?.file || item?.relativeFile),
+    line: normalizeOptionalBatchNumber(item?.line),
+    ref: compactBatchString(item?.ref),
+    marker: compactBatchString(item?.marker),
+    files: compactBatchStringList(item?.files, 8),
+    candidates: depth < 2 ? compactBatchSourceMatches(item?.candidates, 8) : undefined,
+    failures: depth < 2 ? compactBatchDiagnostics(item?.failures, depth + 1) : undefined,
+    checks: depth < 2 ? compactBatchDiagnostics(item?.checks, depth + 1) : undefined,
+  }));
+}
+
+function compactBatchCandidates(candidates) {
+  return (Array.isArray(candidates) ? candidates : [])
+    .slice(0, 24)
+    .map((candidate) => ({
+      entryId: compactBatchString(candidate?.entryId),
+      ref: compactBatchString(candidate?.ref),
+      sourceHint: compactBatchSourceMatch(candidate?.sourceHint),
+      textMatches: compactBatchSourceMatches(candidate?.textMatches, 8),
+      objectKeyMatches: compactBatchSourceMatches(candidate?.objectKeyMatches, 8),
+      contextTextMatches: compactBatchSourceMatches(candidate?.contextTextMatches, 8),
+      locatorMatches: compactBatchSourceMatches(candidate?.locatorMatches, 6),
+    }));
+}
+
+function compactBatchSourceMatches(matches, limit) {
+  if (!Array.isArray(matches)) return undefined;
+  return matches.slice(0, limit).map(compactBatchSourceMatch).filter(Boolean);
+}
+
+function compactBatchSourceMatch(match) {
+  if (!match || typeof match !== 'object') return null;
+  return {
+    file: compactBatchString(match.relativeFile || match.file),
+    line: normalizeBatchNumber(match.line),
+    column: normalizeBatchNumber(match.column),
+    kind: compactBatchString(match.kind),
+    reason: compactBatchString(match.reason || match.kind),
+    status: compactBatchString(match.status),
   };
 }
 
@@ -311,25 +377,77 @@ function compactBatchOp(op) {
     contextRef: op.contextRef,
     tag: op.tag,
     elementId: op.elementId,
-    classes: op.classes,
+    classes: compactBatchStringList(op.classes, 24),
     originalText: op.originalText,
     newText: op.newText,
     deleted: op.deleted === true || undefined,
-    sourceHint: op.sourceHint,
+    sourceHint: normalizeBatchSourceHint(op.sourceHint),
     leaf: compactContextForBatch(op.leaf),
-    nearbyEditableTexts: Array.isArray(op.nearbyEditableTexts) ? op.nearbyEditableTexts.slice(0, 8) : [],
+    nearbyEditableTexts: compactNearbyBatchTexts(op.nearbyEditableTexts),
     container: compactContextForBatch(op.container),
-    contextHints: Array.isArray(op.contextHints) ? op.contextHints.slice(0, 12) : [],
+    contextHints: compactBatchStringList(op.contextHints, 12),
   };
 }
 
+function normalizeBatchSourceHint(hint) {
+  if (!hint || typeof hint !== 'object') return null;
+  let line = normalizeBatchNumber(hint.line);
+  let column = normalizeBatchNumber(hint.column);
+  if ((line === null || column === null) && typeof hint.loc === 'string') {
+    const match = hint.loc.match(/^(\d+)(?::(\d+))?/);
+    if (match) {
+      line = Number(match[1]);
+      if (match[2]) column = Number(match[2]);
+    }
+  }
+  return {
+    file: compactBatchString(hint.file) || '',
+    loc: compactBatchString(hint.loc) || '',
+    line,
+    column,
+  };
+}
+
+function normalizeBatchNumber(value) {
+  if (value === null || value === undefined || value === '') return null;
+  const number = Number(value);
+  return Number.isFinite(number) ? number : null;
+}
+
+function normalizeOptionalBatchNumber(value) {
+  const number = normalizeBatchNumber(value);
+  return number === null ? undefined : number;
+}
+
+function compactNearbyBatchTexts(items) {
+  return (Array.isArray(items) ? items : [])
+    .slice(0, 8)
+    .map((item) => typeof item === 'string' ? { text: truncate(item, BATCH_OP_TEXT_LIMIT) } : {
+      ref: compactBatchString(item?.ref),
+      tag: compactBatchString(item?.tag),
+      classes: compactBatchStringList(item?.classes, 24),
+      text: compactBatchString(item?.text),
+    });
+}
+
+function compactBatchStringList(items, limit) {
+  return (Array.isArray(items) ? items : [])
+    .slice(0, limit)
+    .filter((item) => typeof item === 'string')
+    .map((item) => truncate(item, BATCH_OP_TEXT_LIMIT));
+}
+
+function compactBatchString(value) {
+  return typeof value === 'string' ? truncate(value, BATCH_OP_TEXT_LIMIT) : undefined;
+}
+
 function compactContextForBatch(value) {
   if (!value || typeof value !== 'object') return value || null;
   return {
-    ref: value.ref,
-    tagName: value.tagName,
-    id: value.id,
-    classes: value.classes,
+    ref: compactBatchString(value.ref),
+    tagName: compactBatchString(value.tagName),
+    id: compactBatchString(value.id),
+    classes: compactBatchStringList(value.classes, 24),
     textContent: truncate(value.textContent, 900),
     outerHTML: truncate(stripLiveRuntimeHtml(value.outerHTML), 1800),
   };
@@ -470,12 +588,11 @@ function runClaude(prompt, { cwd, env, resultPath, logPath, timeoutMs = DEFAULT_
   if (env.IMPECCABLE_LIVE_COPY_AGENT_MODEL) {
     args.push('--model', env.IMPECCABLE_LIVE_COPY_AGENT_MODEL);
   }
-  args.push(prompt);
   // Forward env as-is so CLAUDE_CODE_OAUTH_TOKEN and ANTHROPIC_API_KEY flow
   // through. On macOS, `claude /login` stores creds in the Keychain, which a
   // non-TTY subprocess cannot read; setting CLAUDE_CODE_OAUTH_TOKEN (via
   // `claude setup-token`) is the supported headless auth path.
-  return runAgentProcess('claude', args, '', { cwd, env, logPath, timeoutMs, mirrorOutputPath: resultPath });
+  return runAgentProcess('claude', args, prompt, { cwd, env, logPath, timeoutMs, mirrorOutputPath: resultPath });
 }
 
 function runAgentProcess(command, args, stdin, { cwd, env, logPath, timeoutMs, mirrorOutputPath }) {

.agents/skills/impeccable/scripts/live.mjs

@@ -17,7 +17,7 @@
  *   node live.mjs --help
  */
 
-import { execSync } from 'node:child_process';
+import { execFileSync } from 'node:child_process';
 import fs from 'node:fs';
 import path from 'node:path';
 import { fileURLToPath } from 'node:url';
@@ -316,11 +316,17 @@ function globToRegex(pattern) {
 
 function runScript(name, args, options = {}) {
   const scriptPath = path.join(__dirname, name);
-  const cmd = `node "${scriptPath}" ${args.map(a => `"${a}"`).join(' ')}`;
   try {
-    return execSync(cmd, { encoding: 'utf-8', cwd: options.cwd || process.cwd(), timeout: 15_000 });
+    // argv form, never a shell: string interpolation into double quotes would
+    // let a `"` or `$(...)` in any future caller's arg escape into the shell
+    // (issue #476).
+    return execFileSync(process.execPath, [scriptPath, ...args], {
+      encoding: 'utf-8',
+      cwd: options.cwd || process.cwd(),
+      timeout: 15_000,
+    });
   } catch (err) {
-    // execSync throws on non-zero exit; return stdout if any
+    // execFileSync throws on non-zero exit; return stdout if any
     return err.stdout || err.message || '';
   }
 }

.agents/skills/impeccable/scripts/live/browser-script-parts.mjs

@@ -1,6 +1,8 @@
 import fs from 'node:fs';
 import path from 'node:path';
 
+import { LIVE_CHROME_MOUNT_CONTRACT, LIVE_UI_SURFACES } from './ui-surfaces.mjs';
+
 export const LIVE_BROWSER_SCRIPT_PARTS = Object.freeze([
   Object.freeze({ name: 'session-state', file: 'live-browser-session.js' }),
   Object.freeze({ name: 'dom-helpers', file: 'live-browser-dom.js' }),
@@ -32,7 +34,20 @@ export function readLiveBrowserScriptParts(parts, readFile = (filePath) => fs.re
   }));
 }
 
-export function assembleLiveBrowserScript({ token, port, vocabulary, commandPrefix = '/', appRoot = null, parts }) {
+export function assembleLiveBrowserScript({
+  token,
+  port,
+  vocabulary,
+  commandPrefix = '/',
+  appRoot = null,
+  parts,
+  // Defaulted rather than threaded through live-server.mjs: the browser bundle
+  // must always carry the canonical inventory, and a default makes that true by
+  // construction instead of by every caller remembering to pass it. Overridable
+  // so tests can assemble with a stand-in.
+  uiSurfaces = LIVE_UI_SURFACES,
+  mountContract = LIVE_CHROME_MOUNT_CONTRACT,
+}) {
   const prelude =
     `window.__IMPECCABLE_TOKEN__ = '${token}';\n` +
     `window.__IMPECCABLE_PORT__ = ${port};\n` +
@@ -44,7 +59,14 @@ export function assembleLiveBrowserScript({ token, port, vocabulary, commandPref
     `window.__IMPECCABLE_COMMAND_PREFIX__ = ${JSON.stringify(commandPrefix)};\n` +
     // Canonical command vocabulary (values + labels + icons). live-browser.js
     // builds its action picker from this instead of an inline copy.
-    `window.__IMPECCABLE_VOCAB__ = ${JSON.stringify(vocabulary)};\n`;
+    `window.__IMPECCABLE_VOCAB__ = ${JSON.stringify(vocabulary)};\n` +
+    // Canonical Live chrome inventory from live/ui-surfaces.mjs. live-browser.js
+    // is a classic script and cannot import an ES module at runtime, so the list
+    // is serialized here and read off the global there. Node consumers (this
+    // repo's tests, the impeccable-site Live UI lab) import the module directly,
+    // which is what keeps the two from drifting.
+    `window.__IMPECCABLE_LIVE_UI_SURFACES__ = ${JSON.stringify(uiSurfaces)};\n` +
+    `window.__IMPECCABLE_LIVE_MOUNT_CONTRACT__ = ${JSON.stringify(mountContract)};\n`;
 
   const body = parts.map((part) => {
     const file = part.file || path.basename(part.path || '');

.agents/skills/impeccable/scripts/live/ui-surfaces.mjs

@@ -0,0 +1,75 @@
+/**
+ * Canonical inventory of the Live overlay's UI surfaces: one entry per piece of
+ * chrome Live mounts on the user's page, with the element ids that make it up.
+ *
+ * Single source of truth, consumed by:
+ *   - skill/scripts/live/browser-script-parts.mjs — serializes this into
+ *     window.__IMPECCABLE_LIVE_UI_SURFACES__ in the /live.js prelude.
+ *   - skill/scripts/live-browser.js — publishes it on
+ *     window.__IMPECCABLE_LIVE_CHROME_CORE__ for adapters and E2E probes. That
+ *     file is served raw and injected as a classic <script>, so it cannot
+ *     import this module at runtime; it reads the injected global instead, the
+ *     same path live/vocabulary.mjs already takes for the command palette.
+ *   - the private impeccable-site repo — site/components/LiveUiGallery.astro
+ *     and tests/live-ui-lab.test.mjs import LIVE_UI_SURFACES at build time and
+ *     fail the site build when the Live UI lab has no snapshot for a surface
+ *     defined here. That guard only guards if it reads this list rather than a
+ *     copy the site keeps, so this module must stay importable from Node.
+ *     Renaming a key or the module is a breaking change for that build; the
+ *     list was briefly inlined into live-browser.js and the site had to parse
+ *     it back out with a regex.
+ *
+ * Add a surface here and both the browser bundle and the site lab follow.
+ */
+
+/** Id prefix every Live chrome element carries. Mirrored by PREFIX in live-browser.js. */
+export const LIVE_UI_PREFIX = 'impeccable-live';
+
+const id = (suffix) => `${LIVE_UI_PREFIX}-${suffix}`;
+
+/**
+ * The mount contract every Live chrome adapter (DOM, Svelte, ...) satisfies.
+ * Published alongside the surfaces on __IMPECCABLE_LIVE_CHROME_CORE__.
+ */
+export const LIVE_CHROME_MOUNT_CONTRACT = Object.freeze(['root', 'transport', 'state', 'actions']);
+
+export const LIVE_UI_SURFACES = Object.freeze([
+  {
+    key: 'global-bottom-bar',
+    ids: [
+      id('global-bar'), id('global-bar-brand'), id('pick-toggle'), id('insert-toggle'),
+      id('detect-toggle'), id('detect-badge'), id('design-toggle'), id('page-chat'),
+      id('page-chat-input'), id('page-chat-voice'), id('page-chat-send'),
+    ],
+  },
+  { key: 'pending-copy-edit-dock', ids: [id('pending-dock')] },
+  {
+    key: 'element-selection-chrome',
+    ids: [
+      id('highlight'), id('tooltip'), id('bar'), id('selection-pill'), id('input'),
+      id('configure-voice'), id('configure-bar-tooltip'),
+    ],
+  },
+  { key: 'action-picker', ids: [id('picker')] },
+  { key: 'edit-chrome', ids: [id('edit-badge')] },
+  { key: 'generating-row', ids: [id('bar'), id('shader')] },
+  { key: 'variant-cycling-row', ids: [id('bar'), id('params-panel')] },
+  { key: 'variant-params-panel', ids: [id('params-panel')] },
+  { key: 'saving-confirmed-rows', ids: [id('bar')] },
+  {
+    key: 'insert-mode-chrome',
+    ids: [
+      id('insert-line'), id('insert-placeholder'), id('placeholder-resize'), id('insert-input'),
+      id('insert-voice'), id('insert-create'), id('insert-create-tooltip'),
+    ],
+  },
+  { key: 'annotation-chrome', ids: [id('annot'), id('annot-svg'), id('annot-pins'), id('annot-clear')] },
+  { key: 'design-system-panel', ids: [id('design-host')] },
+  { key: 'toasts-and-errors', ids: [id('toast'), id('mount-error')] },
+  { key: 'css-isolation-boundary', ids: [id('root')] },
+].map((surface) => Object.freeze({ ...surface, ids: Object.freeze(surface.ids) })));
+
+/** Every id any surface owns, de-duplicated, in surface order. */
+export const LIVE_UI_COMPONENT_IDS = Object.freeze([
+  ...new Set(LIVE_UI_SURFACES.flatMap((surface) => surface.ids)),
+]);

.agents/skills/impeccable/scripts/pin.mjs

@@ -93,15 +93,17 @@ function commandPrefixForSkillsDir(skillsDir) {
   return CODEX_HARNESSES.has(basename(dirname(skillsDir))) ? '$' : '/';
 }
 
-function generatePinnedSkill(command, metadata, commandPrefix) {
+function generatePinnedSkill(command, metadata, commandPrefix, isCodex) {
   const desc = metadata[command]?.description || `Shortcut for ${commandPrefix}impeccable ${command}.`;
   const hint = metadata[command]?.argumentHint || '[target]';
+  const providerFrontmatter = isCodex
+    ? `metadata:\n  argument-hint: "${hint}"`
+    : `argument-hint: "${hint}"\nuser-invocable: true`;
 
   return `---
 name: ${command}
 description: "${desc}"
-argument-hint: "${hint}"
-user-invocable: true
+${providerFrontmatter}
 ---
 
 ${PIN_MARKER}
@@ -128,7 +130,7 @@ function pin(command, projectRoot) {
 
   for (const skillsDir of harnessDirs) {
     const commandPrefix = commandPrefixForSkillsDir(skillsDir);
-    const content = generatePinnedSkill(command, metadata, commandPrefix);
+    const content = generatePinnedSkill(command, metadata, commandPrefix, commandPrefix === '$');
     // Check if skill already exists (and isn't a pin)
     const skillDir = join(skillsDir, command);
     if (existsSync(skillDir)) {

.agents/skills/impeccable/scripts/serve-question.mjs

@@ -29,25 +29,50 @@
  *       "materials": ["letterpress", "newsprint"],                // optional, rendered as tags
  *       "viewport": "one line: the first-viewport composition",   // optional
  *       "case": "one line: the fusion verdict, honest",           // optional
+ *       "verdict": "competitive", // optional routing tier: "wins" |
+ *                                // "competitive" | "declined". Declined cards
+ *                                // render demoted after the full cards:
+ *                                // narrow, quiet, catalog art as a labeled
+ *                                // thumb, "Adopt anyway" instead of "Build
+ *                                // this". Still choosable; never deleted.
+ *       "kept": "one line: what the direction kept from this declined world",
+ *       "raised": [ { "from": "challenger-x", "raise": "one line" } ],
+ *                                // assigned card only: donations taken from
+ *                                // declined challengers, rendered as named
+ *                                // raise lines under the identity row
  *       "risk": "one line: the honest risk",                      // optional
  *       "body": "fallback prose when the structured fields are absent",
- *       "sketch": ".impeccable/sketches/assigned.webp",  // optional; may not exist
- *                                // yet: the page shimmer-waits and polls the
- *                                // slot until the file lands, so serve first
- *                                // and generate after
+ *       "sketch": ".impeccable/mocks/decision/assigned.webp",  // optional; the card's
+ *                                // full-fidelity direction comp (the field
+ *                                // keeps the sketch era's wire name). May not
+ *                                // exist yet: the page shimmer-waits and
+ *                                // polls the slot until the file lands, so
+ *                                // serve first and generate after
  *       "hero": "https://... or /abs/path.webp",   // optional inspiration image;
  *                                // rides picture-in-picture when a sketch exists
  *       "board": "https://... or /abs/path.webp"   // optional secondary image
  *     }, ...
  *   ],
  *   "reroll": true,          // adds a re-roll action (returns {"optionId":"reroll"})
+ *                            // or { "registers": ["safer", "bolder"] } to add
+ *                            // the register steers beside it: the answer then
+ *                            // carries "register" and the agent re-runs
+ *                            // concept-seed with --register <value>
  *   "canon": true,           // adds the "Play it straight" standing exit;
  *                            // direction rounds only (returns {"optionId":"canon"})
  *   "canonCard": { ... },    // optional: the standing exit as a full card with the
  *                            // same anatomy (label, thesis, palette, sketch, ...);
  *                            // rendered last and visually subordinate. Without it,
  *                            // canon stays a quiet footer action.
- *   "steer": true            // adds a free-text steer field returned with any answer
+ *   "steer": true,           // adds a free-text steer field returned with any answer
+ *   "followup": true         // this round's pick is not terminal: the server
+ *                            // stays open awaiting --update with the next
+ *                            // round (detached mode only), the page shows a
+ *                            // loading hand instead of goodbye, and the
+ *                            // answer carries followup:true so --wait knows
+ *                            // to keep the table. Use it when a decision has
+ *                            // a known second half, e.g. direction first,
+ *                            // then the execution contract.
  * }
  *
  * Options render as large cards: the sketch leads when present, with the
@@ -79,6 +104,7 @@ import fs from 'node:fs';
 import path from 'node:path';
 import { spawn } from 'node:child_process';
 import { fileURLToPath } from 'node:url';
+import { openSystemBrowser } from './lib/open-system-browser.mjs';
 
 function arg(name, fallback = null) {
   const i = process.argv.indexOf(`--${name}`);
@@ -123,11 +149,17 @@ function printAnswer(raw) {
       console.log("CHOSEN CARD: open the chosen world's board and hero images now, before any code. When your harness only reads files, or runs sandboxed, download them INTO the workspace and open the relative path; a sandboxed viewer rejects absolute paths outside it. They set the craft bar the build must reach.");
     }
     if (a.sketch) {
-      console.log('CHOSEN SKETCH: the decision sketch at that path may seed one comp probe; the comp round still renders its full set, because a sketch chose the direction, not the composition.');
+      console.log('CHOSEN COMP: the decision comp at that path is compositional option one. On a comp-led build the comp round adds two variations beside it; on a code-led build it returns at the finish review as the critique reference. Never regenerate it from scratch.');
     }
     if (a.optionId === 'canon') {
       console.log('CANON CHOSEN: the user picked the category standard on purpose. Ask once for two or three products this should sit alongside; their craft level becomes the quality bar. Execute the canon at full commitment, conventions embraced without irony or smuggled quirk.');
     }
+    if (a.optionId === 'reroll' && a.register) {
+      console.log(`REGISTER: the user steered the next hand to the ${a.register} register. Re-run concept-seed with the same key, the next --reroll round, and --register ${a.register}, then follow what it prints; the register is the user's steering, never yours to pre-select.`);
+    }
+    if (a.followup && a.optionId !== 'reroll') {
+      console.log('FOLLOWUP OPEN: the table stays open and the page is showing a loading hand. Deliver the next round now with --update --key <key> --payload <file>, then collect it with --wait; never leave the page waiting on a round you have not sent.');
+    }
   } catch { /* raw answer */ }
 }
 
@@ -143,15 +175,17 @@ if (hasFlag('schema')) {
     title: 'Choose the visual world',
     question: 'The roll assigned Fillmore Handbill. Keep it, take an alternate, or re-roll.',
     options: [
-      { id: 'assigned', label: 'Fillmore Handbill', kicker: 'THE ROLL', lineage: '1966-71 Fillmore psychedelic handbills', thesis: 'The gig poster that treats every release like a one-night stand.', palette: ['#e8452c', '#f5d64c', '#1b2a52', '#f3ead8'], materials: ['letterpress', 'split-fountain ink'], viewport: 'A full-bleed dated bill with the product name in warped display type.', risk: 'Reads nostalgic when the type is set timidly.', sketch: '.impeccable/sketches/assigned.webp', hero: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill-hero.webp', board: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill.webp' },
-      { id: 'challenger-teletext', label: 'Teletext Service', lineage: 'broadcast teletext magazines', thesis: 'The catalog as a broadcast index: pages, not sections.', case: 'Fuses cleanly: releases map to numbered pages.', sketch: '.impeccable/sketches/challenger-teletext.webp', hero: 'https://impeccable.style/worlds/cards/broadcast-programming-teletext-service-hero.webp' },
+      { id: 'assigned', label: 'Fillmore Handbill', kicker: 'THE ROLL', lineage: '1966-71 Fillmore psychedelic handbills', thesis: 'The gig poster that treats every release like a one-night stand.', palette: ['#e8452c', '#f5d64c', '#1b2a52', '#f3ead8'], materials: ['letterpress', 'split-fountain ink'], viewport: 'A full-bleed dated bill with the product name in warped display type.', risk: 'Reads nostalgic when the type is set timidly.', raised: [{ from: 'challenger-microfiche', raise: 'The bill now owns its whole viewport as one continuous printed sheet.' }], sketch: '.impeccable/mocks/decision/assigned.webp', hero: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill-hero.webp', board: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill.webp' },
+      { id: 'model-pick', label: 'The Broadside Ballad', kicker: 'MY PICK', lineage: 'street-sold ballad sheets', thesis: 'Every release printed as the day’s ballad sheet.', palette: ['#1f1c18', '#efe5d0', '#a33327'], materials: ['woodcut', 'rag paper'], viewport: 'One tall sheet, the newest release as today’s ballad.', risk: 'Also the direction most runs in this category land on.', sketch: '.impeccable/mocks/decision/model-pick.webp' },
+      { id: 'challenger-teletext', label: 'Teletext Service', verdict: 'competitive', lineage: 'broadcast teletext magazines', thesis: 'The catalog as a broadcast index: pages, not sections.', palette: ['#0000c0', '#ffff00', '#00c000', '#ffffff'], materials: ['block mosaic', 'phosphor glow'], viewport: 'P100 index page, releases as numbered rows.', case: 'Fuses cleanly: releases map to numbered pages; loses narrowly on clarity.', risk: 'Reads retro-novelty when the grid is not strict.', sketch: '.impeccable/mocks/decision/challenger-teletext.webp', hero: 'https://impeccable.style/worlds/cards/broadcast-programming-teletext-service-hero.webp' },
+      { id: 'challenger-microfiche', label: 'Microfiche Reader', verdict: 'declined', lineage: 'library microfiche stations', palette: ['#101418', '#9fb4c0'], materials: ['film grain', 'backlit glass'], case: 'Fuses poorly: listeners do not identify with archival retrieval.', kept: 'Total environmental commitment.', hero: 'https://impeccable.style/worlds/cards/archives-microfiche-reader-hero.webp' },
     ],
-    reroll: true,
+    reroll: { registers: ['safer', 'bolder'] },
     canon: true,
-    canonCard: { label: 'The category standard', thesis: 'What this category ships, executed impeccably.', viewport: 'The arrangement a visitor expects, at full craft.', sketch: '.impeccable/sketches/canon.webp' },
+    canonCard: { label: 'The category standard', thesis: 'What this category ships, executed impeccably.', palette: ['#ffffff', '#111827', '#2563eb'], materials: ['clean grid', 'product photography'], viewport: 'The arrangement a visitor expects, at full craft.', risk: 'Indistinguishable from the competition by design.', sketch: '.impeccable/mocks/decision/canon.webp' },
     steer: true,
   }, null, 2));
-  console.log('\nOption ids return verbatim in ANSWER; "reroll" and "canon" are reserved. hero/board/sketch accept URLs or local paths; sketch slots may point at files that do not exist yet (serve first, generate after; the page polls until they land, so never block serving on generation). hero on a challenger is the inspiration it draws from and renders picture-in-picture beside the sketch, never as the promise of the build. canonCard renders the standing exit as a subordinate card with the same anatomy; without it, canon stays a quiet footer action. Include canon only for visual-direction rounds; never present it as your own recommendation. Keep thesis and each fact to one short sentence: the card front shows thesis, identity, and a two-line risk, while first viewport and the case read on the card back behind the Details chip, so long facts cost the reader a flip, not the page its scanability. A card with no imagery at all has no back; its full read renders on the front, so a text-only round loses nothing. Sketch aspect follows the surface: portrait at device viewport for native or mobile-first surfaces, landscape otherwise; the page adapts its cards to either.');
+  console.log('\nOption ids return verbatim in ANSWER; "reroll" and "canon" are reserved. hero/board/sketch accept URLs or local paths; sketch slots may point at files that do not exist yet (serve first, generate after; the page polls until they land, so never block serving on generation). hero on a challenger is the inspiration it draws from and renders picture-in-picture beside the sketch, never as the promise of the build. verdict routes rendering: "wins" and "competitive" challengers keep full cards, "declined" ones render demoted after them (narrow, quiet, art as a labeled thumb, "Adopt anyway"), with their kept line on the front; the page reorders declined cards to the end on its own. raised on the assigned card renders each donation as a named raise line. Salience parity: when the assigned card declares no sketch (no image generation this round), catalog art on every card demotes to a labeled thumb, so what looks important is the verdict’s call, never rendering luck. canonCard renders the standing exit as a subordinate card with the same anatomy; without it, canon stays a quiet footer action. Include canon only for visual-direction rounds; never present it as your own recommendation. The pick card is a kicker convention, not a field: kicker "MY PICK" on your top-ranked grounded candidate, one at most, never in the lead slot. Every card gets the full anatomy, challengers, canon, and declined included: thesis, palette, materials, viewport, risk; the seed already hands you each challenger’s system rules, so a card with no palette chips is an authoring gap, not a data gap. Keep thesis and each fact to one short sentence: the card front shows thesis, identity, and a two-line risk, while first viewport and the case read on the card back behind the Details chip, so long facts cost the reader a flip, not the page its scanability. A card with no imagery at all has no back; its full read renders on the front, so a text-only round loses nothing. The sketch slot carries the card’s full-fidelity direction comp (the field keeps its wire name for compatibility). Comp aspect follows the surface: portrait at device viewport for native or mobile-first surfaces, landscape otherwise; the page adapts its cards to either. reroll accepts true or { "registers": ["safer", "bolder"] }: the register buttons steer the next hand along the familiar-to-bold axis, the answer carries "register", and you re-run concept-seed with --register <value> for the next round; offer the registers on direction rounds, and never pre-select one. followup: true keeps the table open after a pick for a second round via --update (direction first, then the execution contract); send the next payload immediately, the page is waiting on it.');
   process.exit(0);
 }
 
@@ -195,12 +229,16 @@ if (hasFlag('wait')) {
   if (!answered()) { console.log(`WAITING: no answer yet after ${pollSec}s; run --wait --key ${key} again`); process.exit(3); }
   const collected = fs.readFileSync(answerFile(key), 'utf8').trim();
   printAnswer(collected);
-  // A re-roll keeps the table open: the server stays alive awaiting --update,
-  // so only the answer file is consumed. Terminal choices clean up fully.
-  let isRerollAnswer = false;
-  try { isRerollAnswer = JSON.parse(collected).optionId === 'reroll'; } catch { /* treat as terminal */ }
+  // A re-roll or a followup-round pick keeps the table open: the server stays
+  // alive awaiting --update, so only the answer file is consumed. Terminal
+  // choices clean up fully.
+  let keepsTableOpen = false;
+  try {
+    const parsedAnswer = JSON.parse(collected);
+    keepsTableOpen = parsedAnswer.optionId === 'reroll' || parsedAnswer.followup === true;
+  } catch { /* treat as terminal */ }
   try { fs.rmSync(answerFile(key)); } catch { /* already gone */ }
-  if (!isRerollAnswer) { try { fs.rmSync(stateFile(key)); } catch { /* already gone */ } }
+  if (!keepsTableOpen) { try { fs.rmSync(stateFile(key)); } catch { /* already gone */ } }
   process.exit(0);
 }
 
@@ -301,11 +339,19 @@ function loadRound(json) {
     sketchSrc: sketchSrc(option.sketch),
   });
   options = parsed.options.map(decorate);
+  // The verdict routes rendering: full cards first, then the canon, then the
+  // declined cards dead last in their own payload order. The reorder happens
+  // here so a payload that interleaves them still renders the weighing's
+  // shape, and the deck reads as a gradient of standing: contenders, the
+  // familiar door, then the demoted row.
+  const declined = options.filter((o) => o.verdict === 'declined');
+  options = options.filter((o) => o.verdict !== 'declined');
   // The standing exit as a full card: same anatomy, reserved id, rendered
   // subordinate by the page. Without it, canon stays the quiet footer action.
   if (parsed.canonCard && typeof parsed.canonCard === 'object') {
     options = [...options, { ...decorate(parsed.canonCard), id: 'canon', isCanon: true }];
   }
+  options = [...options, ...declined];
 }
 try { loadRound(raw); } catch (error) { console.error(`serve-question: ${error.message}`); process.exit(1); }
 const detachedKey = hasFlag('detached-serve') ? arg('key') : null;
@@ -321,7 +367,19 @@ function page() {
   // and material tags give a text-only direction an immediate identity that
   // no generation luck can distort.
   const fact = (label, value, cls = '') => value ? `<p class="fact${cls ? ` ${cls}` : ''}"><span class="fact-label">${label}</span>${esc(value)}</p>` : '';
-  const hasMedia = (option) => Boolean(option.sketchSrc || option.heroSrc || option.boardSrc);
+  const demoted = (option) => option.verdict === 'declined';
+  // Salience parity: a card's imagery weight is capped by the assigned card's.
+  // When the lead card has no media at all (no image generation this round,
+  // and no catalog art of its own), full-bleed catalog art beside a text-only
+  // assigned card would let rendering luck outvote the weighing: users click
+  // the colorful thing. Declined cards are thumb-only regardless; the verdict
+  // demoted them, and a full-bleed hero would promote them right back.
+  const identityRound = !(options[0] && (options[0].sketchSrc || options[0].heroSrc || options[0].boardSrc));
+  // A declined card never renders a full media face, sketch included: even a
+  // declared sketch would buy back the salience the verdict took away.
+  const faceSketch = (option) => demoted(option) ? null : option.sketchSrc;
+  const thumbOnly = (option) => !faceSketch(option) && Boolean(option.heroSrc || option.boardSrc) && (demoted(option) || identityRound);
+  const hasMedia = (option) => Boolean(faceSketch(option) || ((option.heroSrc || option.boardSrc) && !thumbOnly(option)));
   // The back exists to keep long facts off a card whose front is an image;
   // a card with no art has no flip chip to reach it, so it gets no back and
   // the full read lives on the front instead.
@@ -337,6 +395,30 @@ function page() {
       idBits.push(option.materials.slice(0, 4).map((m) => `<span class="tag">${esc(m)}</span>`).join(''));
     }
     if (idBits.length) rows.push(`<div class="identity">${idBits.join('')}</div>`);
+    // Donations from declined challengers render as named raise lines: the
+    // assigned card arrives already raised by the hand it beat, and the raise
+    // is readable, because a raise nobody can read did not happen. One raise
+    // renders inline; several become a compact cycler (click advances), so a
+    // generous hand cannot blow the card out of proportion.
+    if (Array.isArray(option.raised) && option.raised.length) {
+      const nameOf = (id) => options.find((o) => o.id === id)?.label || String(id ?? '');
+      const raiseLines = option.raised.slice(0, 6).map((r) => `<p class="raise"><span class="fact-label">Raised by ${esc(nameOf(r.from))}</span>${esc(r.raise || r.kept || '')}</p>`);
+      if (raiseLines.length > 1) {
+        rows.push(`<div class="raises raises-cycle" role="button" tabindex="0" title="Click or press Enter to see the next raise" aria-label="Raised by the hand; activate to see the next raise">
+              <div class="raises-head"><span class="fact-label">Raised by the hand</span><span class="raises-count" data-raises-count>1/${raiseLines.length}</span></div>
+              ${raiseLines.join('')}
+              <span class="sr-live" aria-live="polite"></span>
+            </div>`);
+      } else {
+        rows.push(`<div class="raises">${raiseLines[0]}</div>`);
+      }
+    }
+    // Demoted art stays reachable as a labeled thumb: the catalog world
+    // explains where the direction comes from without buying it back the
+    // salience the verdict took away.
+    if (thumbOnly(option)) {
+      rows.push(`<figure class="inspo" title="Inspiration: the world this direction draws from. Your page will not look like this image."><img src="${esc(option.heroSrc || option.boardSrc)}" alt=""><figcaption>inspired by</figcaption></figure>`);
+    }
     // The front carries only what the choice needs: thesis, identity, and the
     // honest risk clamped to two lines. First viewport and the case read on
     // the card's back; once the sketch lands, the first viewport is a picture.
@@ -347,6 +429,7 @@ function page() {
     } else {
       rows.push(fact('First viewport', option.viewport));
       rows.push(fact('The case', option.case));
+      rows.push(fact('Kept', option.kept));
       rows.push(fact('Risk', option.risk));
     }
     if (!option.thesis && option.body) rows.push(`<p class="detail">${esc(option.body)}</p>`);
@@ -356,6 +439,7 @@ function page() {
   const backFacts = (option) => [
     fact('First viewport', option.viewport),
     fact('The case', option.case),
+    fact('Kept', option.kept),
     fact('Risk', option.risk),
     option.body && option.thesis ? `<p class="detail more">${esc(option.body)}</p>` : '',
   ].filter(Boolean).join('\n            ');
@@ -365,9 +449,12 @@ function page() {
               <figcaption>inspiration</figcaption>
             </figure>` : '';
     const details = hasBack(option) ? flipChip('Details') : '';
-    if (option.sketchSrc) {
+    // Thumb-only art renders inside the body via anatomy(), never as a face,
+    // and a declined card's sketch slot is ignored outright.
+    if (thumbOnly(option)) return '';
+    if (faceSketch(option)) {
       return `<div class="media sketching" data-sketch="${esc(option.sketchSrc)}">
-            <div class="shimmer"><span class="sketch-note">sketching&hellip;</span></div>
+            <div class="shimmer"><span class="sketch-note">rendering&hellip;</span></div>
             <img class="sketch" alt="" hidden>
             ${inspiration}
             <div class="chips">${expandChip}${details}</div>
@@ -384,17 +471,18 @@ function page() {
     }
     return '';
   };
+  const chooseLabel = (option) => option.isCanon ? 'Play it straight' : demoted(option) ? 'Adopt anyway' : 'Build this';
   const cards = options.map((option, index) => `
-    <article class="card${option.isCanon ? ' canon' : ''}" style="--fan:${index === 0 ? '0deg' : (index % 2 ? '1.4deg' : '-1.2deg')};--deal:${index * 90}ms" data-id="${esc(option.id)}">
+    <article class="card${option.isCanon ? ' canon' : ''}${demoted(option) ? ' declined' : ''}" style="--fan:${index === 0 ? '0deg' : (index % 2 ? '1.4deg' : '-1.2deg')};--deal:${index * 90}ms" data-id="${esc(option.id)}">
       <div class="card-inner">
         <div class="face front${index === 0 ? ' lead' : ''}${media(option) ? '' : ' text-only'}">
-          ${option.kicker ? `<span class="kicker">${esc(option.kicker)}</span>` : option.isCanon ? '<span class="kicker standing">The standing door</span>' : ''}
+          ${option.kicker ? `<span class="kicker">${esc(option.kicker)}</span>` : demoted(option) ? '<span class="kicker declined-k">Declined</span>' : option.isCanon ? '<span class="kicker standing">The standing door</span>' : ''}
           ${media(option)}
           <div class="body">
             ${option.lineage ? `<p class="tier">${esc(option.lineage)}</p>` : ''}
             <h2>${esc(option.label)}</h2>
             ${anatomy(option)}
-            <button class="choose" data-id="${esc(option.id)}">${option.isCanon ? 'Play it straight' : 'Build this'}</button>
+            <button class="choose" data-id="${esc(option.id)}">${chooseLabel(option)}</button>
           </div>
         </div>
         ${hasBack(option) ? `<div class="face back${index === 0 ? ' lead' : ''}">
@@ -405,7 +493,7 @@ function page() {
           <div class="body back-body">
             ${option.boardSrc ? `<p class="tier">The full read &middot; ${esc(option.label)}</p>` : ''}
             ${backFacts(option)}
-            <button class="choose" data-id="${esc(option.id)}">${option.isCanon ? 'Play it straight' : 'Build this'}</button>
+            <button class="choose" data-id="${esc(option.id)}">${chooseLabel(option)}</button>
           </div>
         </div>` : ''}
       </div>
@@ -496,6 +584,10 @@ function page() {
     .nav.next { right: auto; left: 50%; top: auto; bottom: 6px; transform: translate(-50%, 0); }
     .fade-prev { top: 0; left: 0; right: 0; bottom: auto; width: auto; height: 72px; background: linear-gradient(180deg, var(--ks-lacquer), transparent); }
     .fade-next { top: auto; left: 0; right: 0; bottom: 0; width: auto; height: 72px; background: linear-gradient(0deg, var(--ks-lacquer), transparent); }
+    /* In the vertical deck the cross axis is horizontal: flex-start would
+       shrink a declined card to content WIDTH, not height, so it stretches
+       like every other card and its height is already its own. */
+    .grid > .card.declined { align-self: stretch; }
   }
   .card { position: relative; perspective: 1400px; transform: rotate(var(--fan, 0deg)); transition: transform .25s cubic-bezier(.16, 1, .3, 1); }
   .card:hover { transform: rotate(0deg) translateY(-4px); }
@@ -584,6 +676,42 @@ function page() {
   /* The generic .media img display:block would defeat [hidden] and float an
      empty block over the shimmer; an unloaded sketch must truly not render. */
   .media img[hidden] { display: none; }
+  /* Declined challengers: the weighing demoted them, so the card is narrower
+     and quieter, its catalog art rides as a labeled thumb in the body, and
+     the action reads "Adopt anyway". Adoptable, never deleted: the demoted
+     row is the hand's proof of judgment. */
+  /* Narrow AND short: without align-self the stretch default drags a thin
+     declined card to the tallest contender's height, a strange stilt of a
+     card beside the full hand. */
+  .grid > .card.declined { flex: 0 0 clamp(15rem, 21vw, 21rem); align-self: flex-start; }
+  .card.declined .face { background: var(--ks-graphite); }
+  .card.declined:hover .face { border-color: var(--ks-text-faint); }
+  .card.declined h2 { font-size: 1rem; color: var(--ks-text); }
+  .kicker.declined-k { background: transparent; border: 1px solid var(--ks-rule); color: var(--ks-text-faint); }
+  .card.declined button.choose { background: transparent; color: var(--ks-text-muted); border: 1px solid var(--ks-rule); font-size: .85rem; padding: 8px 22px; }
+  .card.declined button.choose:hover { background: var(--ks-graphite-2); border-color: var(--ks-text-muted); }
+  /* Thumb-scale inspiration: present, labeled, zoomable, and incapable of
+     outshouting a text-only assigned card. */
+  .inspo { position: relative; flex: none; margin: 2px 0; width: 104px; height: 64px; border: 1px solid var(--ks-rule); border-radius: 6px; overflow: hidden; cursor: zoom-in; background: var(--ks-lacquer); }
+  .inspo img { display: block; width: 100%; height: 100%; object-fit: cover; }
+  .inspo figcaption { position: absolute; left: 0; right: 0; bottom: 0; font-family: var(--ks-mono); font-size: .48rem; letter-spacing: .16em; text-transform: uppercase; color: var(--ks-text); text-align: center; padding: 2px 0 3px; background: oklch(7% 0.006 95 / 0.72); }
+  /* Raises: the donations the assigned direction took from the hand it beat,
+     each named for its donor. Patina, not kinpaku: a raise is provenance. */
+  .raises { display: flex; flex-direction: column; gap: 4px; margin: 2px 0; }
+  .raise { font-size: .78rem; color: var(--ks-text-muted); line-height: 1.45; border-left: 2px solid var(--ks-patina); padding-left: 8px; }
+  .raise .fact-label { color: var(--ks-patina); }
+  /* Several raises cycle instead of stacking: one visible at a time, a
+     counter for the rest, the whole block advances on click. */
+  .raises-cycle { cursor: pointer; border-radius: 6px; }
+  .raises-cycle .raise { display: none; border-left: none; padding-left: 0; }
+  .raises-cycle .raise.active { display: block; }
+  .raises-cycle { border-left: 2px solid var(--ks-patina); padding-left: 8px; }
+  .raises-head { display: flex; align-items: baseline; justify-content: space-between; gap: 8px; }
+  .raises-head .fact-label { color: var(--ks-patina); }
+  .raises-count { font-family: var(--ks-mono); font-size: .58rem; letter-spacing: .14em; color: var(--ks-text-faint); }
+  .raises-count::after { content: " \\203A"; }
+  .raises-cycle:hover .raises-count { color: var(--ks-patina); }
+  .sr-live { position: absolute; width: 1px; height: 1px; padding: 0; margin: -1px; overflow: hidden; clip: rect(0 0 0 0); white-space: nowrap; border: 0; }
   /* The standing exit as a card: present with full anatomy, never dressed as a
      contender. Graphite instead of kinpaku, and it never takes the lead ring. */
   .card.canon .face { border-color: var(--ks-rule); background: var(--ks-graphite); }
@@ -596,9 +724,14 @@ function page() {
   footer { width: 100%; max-width: 90rem; margin: 1.6rem auto 0; display: flex; gap: 1rem; align-items: center; flex-wrap: wrap; }
   #steer { flex: 1; min-width: 16rem; background: var(--ks-lacquer-raised); color: var(--ks-text); border: 1px solid var(--ks-rule); border-radius: 7px; padding: .6rem .85rem; font: inherit; }
   #steer:focus { outline: none; border-color: var(--ks-patina); }
-  #reroll { display: inline-flex; align-items: center; align-self: stretch; gap: 8px; padding: 0 16px; font-family: var(--ks-mono); font-size: .72rem; letter-spacing: .08em; text-transform: uppercase; color: var(--ks-kinpaku); background: transparent; border: 1px solid var(--ks-rule); border-radius: 6px; cursor: pointer; transition: border-color .2s ease, color .2s ease; }
-  #reroll:hover { color: var(--ks-kinpaku-pale); border-color: var(--ks-kinpaku-deep); }
-  #reroll svg { width: 15px; height: 15px; }
+  .reroll-btn { display: inline-flex; align-items: center; align-self: stretch; gap: 8px; padding: 0 16px; font-family: var(--ks-mono); font-size: .72rem; letter-spacing: .08em; text-transform: uppercase; color: var(--ks-kinpaku); background: transparent; border: 1px solid var(--ks-rule); border-radius: 6px; cursor: pointer; transition: border-color .2s ease, color .2s ease; }
+  .reroll-btn:hover { color: var(--ks-kinpaku-pale); border-color: var(--ks-kinpaku-deep); }
+  .reroll-btn svg { width: 15px; height: 15px; }
+  .reroll-btn[disabled] { opacity: .4; cursor: default; }
+  /* The register steers read quieter than the plain roll: they are exits from
+     the current register, not the round's main verbs. */
+  #reroll-safer, #reroll-bolder { color: var(--ks-text-muted); min-height: 38px; }
+  #reroll-safer:hover, #reroll-bolder:hover { color: var(--ks-text); border-color: var(--ks-text-faint); }
   /* The quiet exit: always available, never argued with, visually subordinate
      to the dealt cards and the re-roll so it reads as the user's own door,
      not a recommendation. */
@@ -643,16 +776,33 @@ function page() {
 </main>
 <footer>
   ${payload.steer ? '<input id="steer" placeholder="Optional steer: what should be different or kept?">' : ''}
-  ${payload.reroll ? '<button id="reroll"><svg viewBox="0 0 24 24" aria-hidden="true"><rect x="3" y="3" width="18" height="18" rx="4" fill="none" stroke="currentColor" stroke-width="1.6"/><circle cx="8.4" cy="8.4" r="1.5" fill="currentColor"/><circle cx="15.6" cy="8.4" r="1.5" fill="currentColor"/><circle cx="8.4" cy="15.6" r="1.5" fill="currentColor"/><circle cx="15.6" cy="15.6" r="1.5" fill="currentColor"/><circle cx="12" cy="12" r="1.5" fill="currentColor"/></svg><span>Re-roll</span></button>' : ''}
+  ${(() => {
+    if (!payload.reroll) return '';
+    const die = '<svg viewBox="0 0 24 24" aria-hidden="true"><rect x="3" y="3" width="18" height="18" rx="4" fill="none" stroke="currentColor" stroke-width="1.6"/><circle cx="8.4" cy="8.4" r="1.5" fill="currentColor"/><circle cx="15.6" cy="8.4" r="1.5" fill="currentColor"/><circle cx="8.4" cy="15.6" r="1.5" fill="currentColor"/><circle cx="15.6" cy="15.6" r="1.5" fill="currentColor"/><circle cx="12" cy="12" r="1.5" fill="currentColor"/></svg>';
+    const registers = Array.isArray(payload.reroll.registers) ? payload.reroll.registers.filter((r) => r === 'safer' || r === 'bolder') : [];
+    // The registers are the user's steering wheel on the familiar-to-bold
+    // axis; the plain re-roll sits between them so the spatial order matches
+    // the axis it names.
+    const safer = registers.includes('safer') ? '<button class="reroll-btn" id="reroll-safer" title="Deal the familiar register: conventional grounded directions plus the category standard against named competitors"><span>&larr; Safer hand</span></button>' : '';
+    const bolder = registers.includes('bolder') ? '<button class="reroll-btn" id="reroll-bolder" title="Deal foreign forms only, at full commitment"><span>Bolder hand &rarr;</span></button>' : '';
+    return `${safer}<button class="reroll-btn" id="reroll">${die}<span>Re-roll</span></button>${bolder}`;
+  })()}
   ${payload.canon && !payload.canonCard ? '<button id="canon" title="Skip the roll: build the page this category ships, executed impeccably">Play it straight</button>' : ''}
 </footer>
 <script>
   const steer = () => document.getElementById('steer')?.value || '';
+  // A followup round's pick keeps the tab: the next round arrives via
+  // --update, so the page shows the loading hand instead of goodbye. Detached
+  // mode only, and the page must agree with the server: a blocking server
+  // exits on any pick and has no update channel, so a followup payload there
+  // still gets the goodbye screen, never a loading hand nothing will resolve.
+  const FOLLOWUP = ${payload.followup === true && Boolean(detachedKey) ? 'true' : 'false'};
   const beat = () => { try { navigator.sendBeacon('/heartbeat'); } catch { fetch('/heartbeat', { method: 'POST' }); } };
   beat();
   setInterval(beat, 5000);
   async function answer(optionId) {
     await fetch('/answer', { method: 'POST', headers: { 'content-type': 'application/json' }, body: JSON.stringify({ optionId, steer: steer() }) });
+    if (FOLLOWUP) { await awaitNextRound(); return; }
     document.body.innerHTML = '<div class="done"><svg viewBox="0 0 24 24" width="38" height="38" fill="oklch(84% 0.19 80.46)" aria-hidden="true"><path d="M5 2.5 L13.5 2.5 L5.5 21.5 L5 21.5 Q2.5 21.5 2.5 19 L2.5 5 Q2.5 2.5 5 2.5 Z"/><path d="M16.5 2.5 L19 2.5 Q21.5 2.5 21.5 5 L21.5 19 Q21.5 21.5 19 21.5 L8.5 21.5 Z"/></svg>Choice recorded. The agent is resuming; you can close this tab.</div>';
   }
   document.querySelectorAll('button.choose').forEach(b => b.addEventListener('click', () => answer(b.dataset.id)));
@@ -661,6 +811,25 @@ function page() {
     b.closest('.card').classList.toggle('flipped');
   }));
 
+  // Raise cycler: click (or Enter) advances to the next donation.
+  document.querySelectorAll('.raises-cycle').forEach(cycle => {
+    const raises = [...cycle.querySelectorAll('.raise')];
+    const count = cycle.querySelector('[data-raises-count]');
+    let at = 0;
+    const live = cycle.querySelector('.sr-live');
+    const show = (announce) => {
+      raises.forEach((raise, i) => raise.classList.toggle('active', i === at));
+      if (count) count.textContent = (at + 1) + '/' + raises.length;
+      // Screen readers hear the raise they just advanced to; the initial
+      // render stays quiet so page load does not narrate every card.
+      if (announce && live) live.textContent = 'Raise ' + (at + 1) + ' of ' + raises.length + ': ' + (raises[at]?.textContent || '');
+    };
+    show(false);
+    const advance = (e) => { e.stopPropagation(); at = (at + 1) % raises.length; show(true); };
+    cycle.addEventListener('click', advance);
+    cycle.addEventListener('keydown', (e) => { if (e.key === 'Enter' || e.key === ' ') { e.preventDefault(); advance(e); } });
+  });
+
   // Deal from the stack: cards begin piled at the grid's center, blurred,
   // then travel to their seats with a stagger.
   const cards = [...document.querySelectorAll('.card')];
@@ -708,7 +877,7 @@ function page() {
     const note = m.querySelector('.sketch-note');
     const started = Date.now();
     // A live elapsed count is the difference between "working" and "frozen".
-    const tick = setInterval(() => { if (note) note.textContent = 'sketching · ' + Math.round((Date.now() - started) / 1000) + 's'; }, 1000);
+    const tick = setInterval(() => { if (note) note.textContent = 'rendering · ' + Math.round((Date.now() - started) / 1000) + 's'; }, 1000);
     const settle = () => { clearInterval(tick); m.classList.remove('sketching', 'stand-in'); m.querySelector('.shimmer')?.remove(); m.querySelector('.stand-in-label')?.remove(); };
     const standIn = () => {
       const pip = m.querySelector('.pip img');
@@ -719,7 +888,7 @@ function page() {
       clearInterval(tick);
       const label = document.createElement('p');
       label.className = 'stand-in-label';
-      label.textContent = 'inspiration · sketch pending';
+      label.textContent = 'inspiration · comp pending';
       m.appendChild(label);
     };
     const tryLoad = () => {
@@ -758,15 +927,15 @@ function page() {
     if (img.complete && img.naturalWidth === 0 && img.getAttribute('src')) artFailed(img);
     else img.addEventListener('error', () => artFailed(img), { once: true });
   });
-  // A broken inspiration PIP just leaves; nothing depends on it.
-  document.querySelectorAll('.pip img').forEach(img => {
-    const gone = () => img.closest('.pip')?.remove();
+  // A broken inspiration PIP or thumb just leaves; nothing depends on it.
+  document.querySelectorAll('.pip img, .inspo img').forEach(img => {
+    const gone = () => img.closest('.pip, .inspo')?.remove();
     if (img.complete && img.naturalWidth === 0) gone();
     else img.addEventListener('error', gone, { once: true });
   });
 
-  // Inspiration PIP opens the full catalog card in the lightbox.
-  document.querySelectorAll('.pip').forEach(p => p.addEventListener('click', (e) => {
+  // Inspiration PIP or body thumb opens the full catalog card in the lightbox.
+  document.querySelectorAll('.pip, .inspo').forEach(p => p.addEventListener('click', (e) => {
     e.stopPropagation();
     const img = p.querySelector('img');
     if (!img) return;
@@ -813,7 +982,7 @@ function page() {
   const ambient = document.getElementById('ambient');
   document.querySelectorAll('.card').forEach(card => {
     card.addEventListener('mouseenter', () => {
-      const art = card.querySelector('.face.front .media img:not([hidden])') || card.querySelector('.face.front .pip img');
+      const art = card.querySelector('.face.front .media img:not([hidden])') || card.querySelector('.face.front .pip img') || card.querySelector('.face.front .inspo img');
       if (!art || !art.getAttribute('src')) return;
       ambient.style.backgroundImage = 'url("' + art.getAttribute('src') + '")'; ambient.style.opacity = '1';
     });
@@ -861,8 +1030,11 @@ function page() {
   lightbox.addEventListener('click', closeLightbox);
   document.addEventListener('keydown', (e) => { if (e.key === 'Escape' && !lightbox.hidden) closeLightbox(); });
   document.getElementById('canon')?.addEventListener('click', () => answer('canon'));
-  document.getElementById('reroll')?.addEventListener('click', async () => {
-    await fetch('/answer', { method: 'POST', headers: { 'content-type': 'application/json' }, body: JSON.stringify({ optionId: 'reroll', steer: steer() }) });
+  const dealAgain = async (register) => {
+    await fetch('/answer', { method: 'POST', headers: { 'content-type': 'application/json' }, body: JSON.stringify({ optionId: 'reroll', steer: steer(), ...(register ? { register } : {}) }) });
+    await awaitNextRound();
+  };
+  async function awaitNextRound() {
     const grid = document.querySelector('.grid');
     const cardsNow = [...grid.querySelectorAll('.card')];
     const g = grid.getBoundingClientRect();
@@ -879,14 +1051,17 @@ function page() {
     }
     const cardHeight = cardsNow[0] ? cardsNow[0].getBoundingClientRect().height : 0;
     grid.innerHTML = cardsNow.map(() => '<article class="card skeleton"' + (cardHeight ? ' style="height:' + cardHeight + 'px"' : '') + '><div class="card-inner"><div class="face front"><div class="media"><div class="shimmer"></div></div><div class="body"><div class="line tier w40"></div><div class="line title w70"></div><div class="line w90"></div><div class="line w80"></div><div class="line w60"></div><div class="line button"></div></div></div></div></article>').join('');
-    document.getElementById('reroll')?.setAttribute('disabled', '');
+    document.querySelectorAll('.reroll-btn').forEach(b => b.setAttribute('disabled', ''));
     const poll = setInterval(async () => {
       try {
         const status = await (await fetch('/next-status')).json();
         if (status.ready) { clearInterval(poll); location.reload(); }
       } catch { /* server briefly busy */ }
     }, 1200);
-  });
+  }
+  document.getElementById('reroll')?.addEventListener('click', () => dealAgain());
+  document.getElementById('reroll-safer')?.addEventListener('click', () => dealAgain('safer'));
+  document.getElementById('reroll-bolder')?.addEventListener('click', () => dealAgain('bolder'));
 </script>`;
 }
 
@@ -943,22 +1118,29 @@ const server = http.createServer((req, res) => {
       let parsed = {};
       try { parsed = JSON.parse(body); } catch { /* empty steer */ }
       const chosen = options.find((o) => o.id === parsed.optionId);
+      const isReroll = parsed.optionId === 'reroll';
+      // A followup round's pick is not terminal: the table stays open for the
+      // next round (--update), exactly like a re-roll. Detached mode only;
+      // the blocking mode has no update channel, so its picks stay terminal.
+      const followupOpen = Boolean(detachedKey) && payload.followup === true && !isReroll;
       const answer = JSON.stringify({
         optionId: parsed.optionId ?? null,
         steer: parsed.steer ?? '',
+        ...(isReroll && (parsed.register === 'safer' || parsed.register === 'bolder') ? { register: parsed.register } : {}),
+        ...(followupOpen ? { followup: true } : {}),
         ...(chosen?.hero || chosen?.board ? { hero: chosen.hero ?? null, board: chosen.board ?? null } : {}),
         ...(chosen?.sketch ? { sketch: chosen.sketch } : {}),
       });
-      const isReroll = parsed.optionId === 'reroll';
       if (detachedKey) {
         fs.mkdirSync(QUESTION_DIR, { recursive: true });
         fs.writeFileSync(answerFile(detachedKey), answer + '\n');
       } else {
         printAnswer(answer);
       }
-      // A re-roll in detached mode keeps the table open: the client shows a
-      // loading hand and reloads when --update delivers the next round.
-      if (!(isReroll && detachedKey)) setTimeout(() => process.exit(0), 150);
+      // A re-roll or followup pick in detached mode keeps the table open: the
+      // client shows a loading hand and reloads when --update delivers the
+      // next round.
+      if (!((isReroll || followupOpen) && detachedKey)) setTimeout(() => process.exit(0), 150);
     });
     return;
   }
@@ -976,8 +1158,7 @@ server.listen(portArg, '127.0.0.1', () => {
     console.log('Waiting for the user to choose in the browser (Ctrl-C aborts)...');
   }
   if (!hasFlag('no-open')) {
-    const opener = process.platform === 'darwin' ? 'open' : process.platform === 'win32' ? 'start' : 'xdg-open';
-    try { spawn(opener, [url], { stdio: 'ignore', detached: true }).unref(); } catch { /* URL printed anyway */ }
+    openSystemBrowser(url);
   }
   if (timeoutSec > 0) {
     setTimeout(() => {

.claude/agents/impeccable-asset-producer.md

@@ -16,9 +16,9 @@ Your job is production cleanup, not new art direction. Work only from the approv
 
 Do not redesign. Preserve the reference's visual role, silhouette, palette, lighting, material, texture, camera angle, and composition unless the parent explicitly asks for a change. Preserve perspective only when it belongs to the object or scene itself; if CSS should create the card transform, shadow, rounded clipping, border, or layout, remove that presentation chrome from the raster.
 
-## Decision Sketches
+## Decision Comps
 
-When the parent hands you a decision card packet instead of an approved mock, the job is one sketch: one card, one file, written to the card's declared `sketch` path the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a sketch is reported back, not padded from imagination. Render through the parent's shared frame, including its aspect: the requested surface's first viewport as a flat, matte design sketch in the card's own palette and type character, deliberately unfinished, no photorealism, no gloss; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. The frame is shared across siblings so no sketch looks more finished than another; a finish gap breaks the comparison. The only legible text is the product's real name and one real headline; greek every other text region into indistinct lines, because an invented spec, price, or date in a sketch is a claim PRODUCT.md never made. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a sketch run.
+When the parent hands you a decision card packet instead of an approved mock, the job is one comp: one card, one file, written to the card's declared `sketch` path (the field keeps its wire name) the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a comp is reported back, not padded from imagination. Render the card's direction as a north-star comp at full fidelity: the requested surface's first viewport, prompt led by the surface's own structure (its regions named in order with their scale relationships, never the world's atmosphere), fully committed in the card's own palette, type character, and material world; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. Every sibling renders at the same full fidelity in its own grammar, one surface, one aspect; equal commitment is what keeps the comparison honest. Real product name and real content only; never invent commercial claims, prices, benchmarks, or dates PRODUCT.md does not carry. Write the prompt sidecar beside the file. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a decision-comp run.
 
 ## Input Contract
 

.claude/agents/impeccable-finish-reviewer.md

@@ -16,12 +16,12 @@ A hard turn ceiling ends the run without warning; a run that ends before the fiv
 
 ## Input Contract
 
-Expect: the original request; the confirmed user answers; the artifact path(s); desktop and mobile screenshot paths captured by the parent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and the approved comp path; and the skill's `reference/craft-floor.md` path. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
+Expect: the original request; the confirmed user answers; the artifact path(s); the screenshots the parent captured, which live in `.impeccable/review/` (on the web, `desktop.png` and `mobile.png`; on native, device-class names such as `phone.png` and `tablet.png`, suffixed per OS on adaptive); a screenshot path the calling brief names is authoritative when the file exists, and `.impeccable/review/` is where to look when the brief names none or a named path is missing, never a filename you invent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and, on a comp-led build, the approved comp path (a code-led build has no approved comp; it passes the chosen decision comp as a separate critique-reference input, labeled as such, and nothing in this file that binds “the approved comp” binds it); and the skill's `reference/craft-floor.md` path. On a native (`ios` / `android` / `adaptive`) build the packet also carries the platform reference path(s) (`reference/ios.md` / `reference/android.md`) and a line saying no detector ran: read the platform reference alongside the craft floor and judge every check in the platform's own conventions, the screenshots are device captures rather than browser viewports, and your floor check is the build's only slop gate. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
 
 ## Checks, in order
 
-1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comps with no recorded pick mean the approval point was skipped, and that is a material finding.
-2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
+1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comp-round comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comp-round comps with no recorded pick mean the approval point was skipped, and that is a material finding. Files under `.impeccable/mocks/decision/` are exempt: they are the direction round's dealt hand, produced before any comp round, and they imply no approval whatever the build path; a code-led build has no comp round at all.
+2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. A critique-reference comp, when one arrived on such a build, is provocation rather than spec: no element matrix, no adaptation citations, no asset obligations; its one contribution is the question of what the image dared that the build did not, and the dares worth adopting enter material_fixes as ordinary ordered fixes. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
 3. **Ceiling.** Against the QUALITY BAR card: name the world's native devices the build left unused, frame, depth, lettering treatment, ornament density, motion. The card governs commitment and finish, never composition.
 4. **Contract, promise by promise.** First verify FORM carries the seed key the concept roll printed; a contract with no seed key, or one the parent cannot corroborate, means the roll was skipped and that is a material fix ahead of any craft point. Then, for each of the five blocks, does the render keep the promise? Apply the memory test to the first viewport.
 5. **Truth.** Demonstration data authored and labeled synthetic; no invented commercial claims; unanswered claims present as marked placeholders, not omissions. Every image-native region of the approved comp shipped as a real asset, not a gradient standing in for one, and every produced asset visibly present in the screenshots; an asset applied at near-zero opacity or buried behind other paint is a compliance token, not a shipped material.
@@ -39,4 +39,4 @@ Return the disposition line first, then exactly five sections: `persistence` (pa
 
 ## Verdict Pass
 
-When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.
+When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent recaptures over the same screenshot files you read in the review round, so re-read those exact paths for this round; a round-stamped filename you invent points at nothing. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.

.claude/skills/impeccable/SKILL.md

@@ -15,11 +15,11 @@ This skill gives you the tools and permission to create design that earns to be
 Core principles:
 - Go all out. No hedging, no shortcuts. The deliverable must be complete (except assets the user must provide).
 - Dream big and bold. Distinct, beautiful, outstanding and highly inspiring work.
-- Verify in bounded passes, not a loop, and the ceiling covers the whole cycle: screenshots, defect scans, micro-edits, and rebuilds alike. Build fully, inspect once with a batched round (desktop and mobile together), fix everything it shows in one batch, confirm with at most one more round, and stop polishing. Open-ended self-QA burns the user's money doing worse what the finish handoffs do better.
+- Verify in bounded passes, not a loop, and the ceiling covers the whole cycle: screenshots, defect scans, micro-edits, and rebuilds alike. Build fully, inspect once with a batched round (desktop and mobile together on the web; the shipped device classes on a native platform), fix everything it shows in one batch, confirm with at most one more round, and stop polishing. Open-ended self-QA burns the user's money doing worse what the finish handoffs do better.
 
 ## Setup
 
-1. Run `node .claude/skills/impeccable/scripts/context.mjs` once per session (if the runtime shows this skill's loaded base directory, run `node <skill-base-dir>/scripts/context.mjs`; keep cwd at the user's project). Pass a named source file or route as `--target <path>`. It loads PRODUCT.md, DESIGN.md, the matching surface brief, and native-platform guidance when applicable; follow its directives and do not rerun it.
+1. Run `node <skill-base-dir>/scripts/context.mjs` once per session, where `<skill-base-dir>` is the loaded base directory the runtime reports for this skill; keep cwd at the user's project. That base directory resolves every `node .claude/skills/impeccable/scripts/...` command in this skill and its references, and `.claude/skills/impeccable/scripts` is the fallback only when the runtime reports no base directory. Pass a named source file or route as `--target <path>`. It loads PRODUCT.md, DESIGN.md, the matching surface brief, and native-platform guidance when applicable; follow its directives and do not rerun it.
 2. Before acting, load the one playbook that owns the request: the Commands table's reference for an explicit or clearly implied sub-command, or [reference/new-work.md](reference/new-work.md) for a new surface or replacement visual world. Then inspect the target and at least one representative source of incumbent visual truth (tokens, theme, CSS, component, or asset) before editing.
 3. After analysis and direction are resolved, load [reference/craft-floor.md](reference/craft-floor.md) immediately before editing UI. It carries the quality floor, the absolute bans, and the reflexes no detector catches. Do not load it for planning-only work.
 

.claude/skills/impeccable/reference/android.md

@@ -38,3 +38,9 @@ Would a fluent Android user trust this app, or trip on off-spec components? The
 - **One FAB, one primary action.** Never stack FABs or spend one on a secondary task.
 - **Snackbars for transient feedback** (actionable when useful, never a toast for that); dialogs only for decisions that must interrupt.
 - **Material motion patterns.** Container transform, shared-axis, fade-through, with standard easing and durations; honor the system Remove animations setting with a crossfade or instant cut.
+
+## Verifying the build
+
+- **Screenshots come from the emulator or a connected device, never a browser.** Build and install, then capture with `adb exec-out screencap -p > <path>` (pick a device with `adb -s <serial>` when several are attached). Capture every device class the app ships to, at least one phone and, when tablets are a target, one tablet, and write the files where the review flow expects them.
+- **Dark theme and font scale belong in the pass.** `adb shell cmd uimode night yes` flips the theme; `adb shell settings put system font_scale 1.3` (restore `1.0` after) catches the clipped labels a fixed layout hides; with several targets attached, the capture's `-s <serial>` goes on these commands too.
+- **Emulators give breadth; gestures, refresh rates, and performance need hardware.** Say which one produced the evidence.

.claude/skills/impeccable/reference/animate.md

@@ -74,12 +74,15 @@ Keep content visible in the default state so failed scripts do not hide the page
 
 Respect autoplay and sound preferences. Any nonessential loop must stop when offscreen or hidden.
 
+Every web animation needs a `prefers-reduced-motion` path with an intentional alternative. Remove or reduce spatial movement while preserving opacity, color, and state transitions that carry meaning. Reduced motion means fewer and gentler animations, not disabling all motion; feedback that confirms an action should remain legible.
+
 ## Verify
 
 - The focal motion is specific to the selected world and surface.
 - Every supporting animation explains feedback, state, or relationship.
 - Interruption and repeated use behave correctly.
 - Desktop, mobile, and keyboard paths remain usable.
+- The `prefers-reduced-motion` path reduces movement without erasing meaningful feedback or state changes.
 - Expensive effects stay smooth on the target device.
 - Removing an animation would lose meaning or authored character, not merely decoration.
 

.claude/skills/impeccable/reference/bolder.md

@@ -1,5 +1,7 @@
 > **Additional context needed**: which section is the target, and what must stay untouched.
 
+An open direction round owns the word first: "bolder" said while a direction decision is on the table is the Bolder hand register steer, a fresh deal of foreign forms (see new-work.md), not this command. This command refines a surface whose world already shipped.
+
 "Bolder" is an amplification request, and almost always it is scoped to something that already exists. The surrounding page, its system, and its conventions are the given. Your job is to raise one part to the conviction the rest already implies, without rebuilding anything the brief did not name. The reflex answer, reaching for more effects, is the opposite of bold; reject it first.
 
 ## Scope is sovereign

.claude/skills/impeccable/reference/craft-floor.md

@@ -12,6 +12,7 @@ Each of these is a check on the built result, not an intention. Run them togethe
 - **Type:** body measure 65–75ch, display max 6rem, tracking floor -0.04em, balanced headings, obvious scale and weight steps. Run the real copy at every breakpoint and fix what overflows.
 - **Motion:** one authored moment, not scattered effects and not one identical entrance on every section. Exponential ease-out from an already-visible default. Reach past transform and opacity: blur, backdrop-filter, clip-path, mask, and shadow belong to the palette when they stay smooth.
 - **States:** hover, disabled, loading, error, empty. Plus real content, working controls, responsive composition, keyboard focus.
+- **Browser surfaces:** the parts you did not draw still carry the design. Text selection, the caret, custom scrollbars, focus rings, underline offset, and the numerals in tabular data all ship with browser defaults that belong to no design system. Theme them from the palette. This is the cheapest signal that a page was built rather than assembled, and the one models skip most reliably.
 - **Copy:** the product's own language. Controls name their action; errors name the problem and the recovery.
 - **Coverage:** every brief requirement present and findable within seconds.
 

.claude/skills/impeccable/reference/degraded/asset-producer.md

@@ -11,9 +11,9 @@ Your job is production cleanup, not new art direction. Work only from the approv
 
 Do not redesign. Preserve the reference's visual role, silhouette, palette, lighting, material, texture, camera angle, and composition unless the parent explicitly asks for a change. Preserve perspective only when it belongs to the object or scene itself; if CSS should create the card transform, shadow, rounded clipping, border, or layout, remove that presentation chrome from the raster.
 
-## Decision Sketches
+## Decision Comps
 
-When the parent hands you a decision card packet instead of an approved mock, the job is one sketch: one card, one file, written to the card's declared `sketch` path the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a sketch is reported back, not padded from imagination. Render through the parent's shared frame, including its aspect: the requested surface's first viewport as a flat, matte design sketch in the card's own palette and type character, deliberately unfinished, no photorealism, no gloss; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. The frame is shared across siblings so no sketch looks more finished than another; a finish gap breaks the comparison. The only legible text is the product's real name and one real headline; greek every other text region into indistinct lines, because an invented spec, price, or date in a sketch is a claim PRODUCT.md never made. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a sketch run.
+When the parent hands you a decision card packet instead of an approved mock, the job is one comp: one card, one file, written to the card's declared `sketch` path (the field keeps its wire name) the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a comp is reported back, not padded from imagination. Render the card's direction as a north-star comp at full fidelity: the requested surface's first viewport, prompt led by the surface's own structure (its regions named in order with their scale relationships, never the world's atmosphere), fully committed in the card's own palette, type character, and material world; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. Every sibling renders at the same full fidelity in its own grammar, one surface, one aspect; equal commitment is what keeps the comparison honest. Real product name and real content only; never invent commercial claims, prices, benchmarks, or dates PRODUCT.md does not carry. Write the prompt sidecar beside the file. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a decision-comp run.
 
 ## Input Contract
 

.claude/skills/impeccable/reference/degraded/finish-reviewer.md

@@ -11,12 +11,12 @@ A hard turn ceiling ends the run without warning; a run that ends before the fiv
 
 ## Input Contract
 
-Expect: the original request; the confirmed user answers; the artifact path(s); desktop and mobile screenshot paths captured by the parent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and the approved comp path; and the skill's `reference/craft-floor.md` path. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
+Expect: the original request; the confirmed user answers; the artifact path(s); the screenshots the parent captured, which live in `.impeccable/review/` (on the web, `desktop.png` and `mobile.png`; on native, device-class names such as `phone.png` and `tablet.png`, suffixed per OS on adaptive); a screenshot path the calling brief names is authoritative when the file exists, and `.impeccable/review/` is where to look when the brief names none or a named path is missing, never a filename you invent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and, on a comp-led build, the approved comp path (a code-led build has no approved comp; it passes the chosen decision comp as a separate critique-reference input, labeled as such, and nothing in this file that binds “the approved comp” binds it); and the skill's `reference/craft-floor.md` path. On a native (`ios` / `android` / `adaptive`) build the packet also carries the platform reference path(s) (`reference/ios.md` / `reference/android.md`) and a line saying no detector ran: read the platform reference alongside the craft floor and judge every check in the platform's own conventions, the screenshots are device captures rather than browser viewports, and your floor check is the build's only slop gate. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
 
 ## Checks, in order
 
-1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comps with no recorded pick mean the approval point was skipped, and that is a material finding.
-2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
+1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comp-round comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comp-round comps with no recorded pick mean the approval point was skipped, and that is a material finding. Files under `.impeccable/mocks/decision/` are exempt: they are the direction round's dealt hand, produced before any comp round, and they imply no approval whatever the build path; a code-led build has no comp round at all.
+2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. A critique-reference comp, when one arrived on such a build, is provocation rather than spec: no element matrix, no adaptation citations, no asset obligations; its one contribution is the question of what the image dared that the build did not, and the dares worth adopting enter material_fixes as ordinary ordered fixes. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
 3. **Ceiling.** Against the QUALITY BAR card: name the world's native devices the build left unused, frame, depth, lettering treatment, ornament density, motion. The card governs commitment and finish, never composition.
 4. **Contract, promise by promise.** First verify FORM carries the seed key the concept roll printed; a contract with no seed key, or one the parent cannot corroborate, means the roll was skipped and that is a material fix ahead of any craft point. Then, for each of the five blocks, does the render keep the promise? Apply the memory test to the first viewport.
 5. **Truth.** Demonstration data authored and labeled synthetic; no invented commercial claims; unanswered claims present as marked placeholders, not omissions. Every image-native region of the approved comp shipped as a real asset, not a gradient standing in for one, and every produced asset visibly present in the screenshots; an asset applied at near-zero opacity or buried behind other paint is a compliance token, not a shipped material.
@@ -34,4 +34,4 @@ Return the disposition line first, then exactly five sections: `persistence` (pa
 
 ## Verdict Pass
 
-When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.
+When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent recaptures over the same screenshot files you read in the review round, so re-read those exact paths for this round; a round-stamped filename you invent points at nothing. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.

.claude/skills/impeccable/reference/hooks.md

@@ -48,14 +48,20 @@ The first argument is the action. Defaults to `status`.
 5. If `<action>` is `ignore-value`, `ignore-file`, or `ignore-rule`, just print the script output. The default scope is shared `.impeccable/config.json`; add `--local` only when the user explicitly asks for a private exception.
 6. If `<action>` is `status`, just print the script output. Do not add commentary unless the user asked a follow-up question.
 
-## Intentional findings
+## Triage findings
 
-The hook itself never writes ignore config. Persist an exception only after the user explicitly confirms the flagged issue is intentional, and always go through `hook-admin.mjs`.
+The hook itself never writes ignore config; every exception goes through `hook-admin.mjs`. Triage each finding into one of three outcomes:
+
+- **Real design problem**: fix it. Never add an ignore to skip a fix or to push a blocked write through.
+- **Confident false positive or sanctioned exception**: persist the narrowest ignore yourself and disclose it in your reply. The bar is evidence you can name: an intentional demo or fixture, documentation of bad design, literal or domain-appropriate motion (a ball that bounces), or a choice the user already confirmed. Put that evidence in `--reason` as `"<who decided: evidence>"`; write "user confirmed" only when the user actually did.
+- **Unsure**: leave the finding standing and ask the user in one line. Ask once; a one-line question costs less than the hook re-firing on every later edit.
+
+Self-serve stops at `ignore-value`. `ignore-file` and `ignore-rule` silence too much to add on your own judgment; ask the user first.
 
 Prefer the narrowest exception:
 
-- If the finding line shows an exact `ignore-value` command, run that command. This writes shared `.impeccable/config.json` by default.
-- For value-specific findings such as `overused-font` and `bounce-easing`, use `ignore-value` when the user confirms the specific value. Do not use `ignore-rule overused-font` for a specific font.
+- If the finding line shows an `ignore-value <rule> <value>` pair, pass it to `hook-admin.mjs ignore-value` with your `--reason`. This writes shared `.impeccable/config.json` by default.
+- For value-specific findings such as `overused-font` and `bounce-easing`, use `ignore-value` for the specific value. Do not use `ignore-rule overused-font` for a specific font.
 - If the finding has no value-specific command, such as `side-tab`, scope that one rule to the file: `ignore-value <id> "*" --file <path>`. Run `npx impeccable detect <path>` first to see what actually fires there.
 - Reach for `ignore-file <path>` only when the whole file is out of scope for design review: a fixture, a generated artifact, a deliberate slop demo. It silences every rule for that file permanently, including rules that have not been written yet. A real UI surface with one noisy rule wants the file-scoped value ignore above.
 - Use `ignore-rule <id>` only when the user asks to suppress that whole rule across the project. For broad overused-font suppression, use `ignore-rule overused-font --all-values` only when the user asks to ignore overused fonts generally.
@@ -67,10 +73,10 @@ Example value-specific exception:
 node .claude/skills/impeccable/scripts/hook-admin.mjs ignore-value overused-font Inter --shared --reason "User confirmed Inter is intentional"
 ```
 
-Example intentional motion exception:
+Example self-served exception, with the evidence named:
 
 ```bash
-node .claude/skills/impeccable/scripts/hook-admin.mjs ignore-value bounce-easing bounce-ball --shared --reason "User confirmed ball bounce animation is intentional"
+node .claude/skills/impeccable/scripts/hook-admin.mjs ignore-value bounce-easing bounce-ball --shared --reason "Agent: literal ball-bounce animation, bounce easing is the subject"
 ```
 
 Example whole-rule font exception:

.claude/skills/impeccable/reference/ios.md

@@ -43,3 +43,9 @@ Would a fluent iPhone user trust this app, or pause at off-spec controls? The te
 
 - **System transitions.** Push slides, sheets rise, dismiss reverses the entrance. Custom transitions that fight the navigation model disorient.
 - **Honor Reduce Motion.** Crossfade instead of parallax and large slides.
+
+## Verifying the build
+
+- **Screenshots come from the Simulator, never a browser.** Build and run, then capture with `xcrun simctl io booted screenshot <path>` (with several running, replace `booted` with the target's UDID from `xcrun simctl list devices booted`; display names can collide, the UDID never does). Capture every device class the app ships to, at least one iPhone and, when iPad is a target, one iPad, and write the files where the review flow expects them.
+- **Dark Mode and Dynamic Type belong in the pass.** `xcrun simctl ui booted appearance dark` flips appearance, reusing the capture's UDID when several are booted; a check at a large Dynamic Type size catches the truncation a fixed layout hides.
+- **Simulators give breadth; posture, gestures, and performance need hardware.** Say which one produced the evidence.

.claude/skills/impeccable/reference/new-work.md

@@ -43,12 +43,14 @@ The script assigns which structure gets built; your top-ranked structure is what
 1. Name the product's unique mechanism in one sentence, the audience's real scene, its cultural home, and what this first surface must prove. Note the page this category always ships and its predictable opposite; name both as the rut and keep them out of the seven-candidate list. A brief that paints its own picture, a product name, a titled artifact, a governing metaphor, adds its literal reading to the rut: spend at most one candidate on it and derive the rest from elsewhere in the audience's world.
 2. From that cultural world, list seven concrete visual systems, artifacts, places, or rituals the audience knows by heart, each with one line on why it resonates and can carry the mechanism, ordered by resonance. The audience's world includes its graphic and screen traditions, not only its physical objects: the notation, publications, identity programs, data graphics, and interfaces it reads daily; a nameable abstract system (a school of poster, a documentation standard) is as concrete a candidate as any artifact. What would this thing look like as a physical object; what did its world look like before the web? Near-duplicates count once. When more than three of the seven share one material family, the derivation stopped at the subject's most obvious artifact; dig until the list spans at least three families.
 3. Turn that material into complete directions: each joins a reusable visual world to a concrete first-surface experience.
-4. Run `node .claude/skills/impeccable/scripts/concept-seed.mjs --scope direction --mode <mode>` and follow what it prints. This step has no substitute and no skip condition: on a new or replacement world, writing artifact code before this script has run and its assignment is acknowledged is a contract violation, whatever the harness, the model, or the time pressure, because the roll is the mechanism that keeps every run from converging on the category default. The script assigns which direction gets built and deals catalog challengers. Fuse each challenger before judging it: the challenger supplies the form and its system grammar, the product supplies every fact, and clarity wins conflicts. Weigh fused challengers against the assigned direction on exactly two axes, audience identification and product clarity; losing to strong grounded material is a valid outcome, and beating a thin or tool-monoculture list is the point.
-5. Present one direction, fully committed: its world, first viewport, visitor path, signature interaction, cross-surface reach, and honest risk. Alongside it, offer the hand's challengers as named alternates, the weighing's verdict written on each as its one-line case, an honest "fuses poorly because X" included; the weighing informs the user's choice, it never pre-empts it. A hand holds at most three challengers: when the roll deals more, the three strongest join the hand and the rest wait in the re-roll pool, noted in one line; dropping a challenger from the hand itself takes a named product-truth failure, disclosed. Add re-roll with an optional one-line steer. Never present a ranked menu of your own grounded candidates; a lineup of those invites the safest card. The two channels share this structure and differ only in richness: cards and boards on the decision page, names and one-liners through the structured tool; the structured tool's option list also carries the standing exit as its last option.
+4. Run `node .claude/skills/impeccable/scripts/concept-seed.mjs --scope direction --mode <mode>` and follow what it prints. This step has no substitute and no skip condition: on a new or replacement world, writing artifact code before this script has run and its assignment is acknowledged is a contract violation, whatever the harness, the model, or the time pressure, because the roll is the mechanism that keeps every run from converging on the category default. The script assigns which direction gets built and deals catalog challengers. Fuse each challenger before judging it: the challenger supplies the form and its system grammar, the product supplies every fact, and clarity wins conflicts. Weigh fused challengers against the assigned direction on exactly two axes, audience identification and product clarity; losing to strong grounded material is a valid outcome, and beating a thin or tool-monoculture list is the point. The weighing closes with a verdict per challenger, decided before any borrowing is considered: wins (beats the assigned direction on both axes; it becomes the build candidate), competitive (holds one axis; it stays a full alternate), or declined (loses both). A declined challenger is not spent: name the one discipline of its system the assigned direction lacks, and raise the assigned direction to match before presenting it. A donation transfers ambition and system discipline (a palette's total commitment, a grid's density courage, a form's structural honesty), never the challenger's clothes; a motif lifted from a declined world is a costume note, not a raise, and one world owns the page. Write each raise into the presented direction as its own line, named for its donor; a raise nobody can read did not happen.
+5. Present one direction, fully committed and already raised by the hand it beat, its raises visible as named lines: its world, first viewport, visitor path, signature interaction, cross-surface reach, and honest risk. Alongside it, route each dealt challenger by its verdict: winning and competitive challengers are full alternates carrying their QUALITY BAR cards and one-line case, while declined challengers render demoted, compact and quiet, each carrying its verdict plus what the direction kept from it, never full-size and never silently dropped, each still adoptable on request. The verdict informs the user's choice, it never pre-empts it; the demoted row is the hand's proof of judgment, showing why the dealt worlds made the presented direction better. A hand holds at most three full-card challengers: when the roll deals more, the three strongest join the hand and the rest wait in the re-roll pool, noted in one line; dropping a challenger from the hand itself takes a named product-truth failure, disclosed. Add one card for your own top-ranked grounded candidate when it is not the assigned direction, kicker MY PICK, same anatomy as every card, with an honest risk line naming its familiarity when true: the strongest grounded direction is often the one most runs in this category land on, and the user deciding that trade is the point of showing it. Familiar and effective is a legitimate destination, not a failure of nerve; the pick card and the standing exit serve it at two depths. One pick card, never two, never a ranked list: the rest of your grounded candidates stay yours, because a lineup of them hands selection back to a taste function and invites the safest card. The pick never takes the lead position, and when the dice assign your top candidate there is no pick card; the assigned card notes it also topped your list. Add re-roll with an optional one-line steer, offered in three registers: plain (a fresh hand, same spread), safer (the familiar register: your remaining conventional grounded candidates plus the canon against named competitors), and bolder (foreign forms only, at full commitment). A register is the user's steering on the familiar-to-bold axis, never yours to pre-select; when the answer carries one, re-run the seed with `--register <value>` and the next `--reroll` round, and follow what it prints. A user saying "bolder" or "safer" while a direction round is open means these registers, never the bolder or harden commands. The two channels share this structure and differ only in richness: cards and boards on the decision page, names and one-liners through the structured tool; the structured tool's option list carries the assigned direction, the pick, the winning and competitive challengers, and the standing exit as its last option, while declined challengers fold into the assigned option's description as their kept lines, so the raise survives the text channel too.
 
-The standing exit: every direction round offers one quiet, permanent alternative, the category standard, played straight. It is the user's door, never yours: never recommend it, never weigh it against the roll, never let it soften the dealt directions; the counterweights bind the unchosen default, not the chosen one. When the user takes it, in the canon action, a safer-steer, or plain words asking for the familiar or competitor-like path, convention becomes the commitment: ask once for two or three products this should sit alongside, make their craft level the bar, and execute the canon at full fidelity, without irony or smuggled quirk. A standing preference gets recorded as a brand commitment in PRODUCT.md. Re-roll eliminates every direction already shown, grounded and challenger alike; after two consecutive re-rolls, ask what quality is missing. You may re-roll on your own only on named factual grounds, when the assigned direction cannot carry the product's truth or task; taste is never grounds. The user may re-roll freely, and a user- or brief-pinned direction beats the roll, always. Present the decision visually: write an options payload with the assigned direction leading, the dealt challengers as alternates carrying their QUALITY BAR cards, and re-roll, steer, plus canon enabled; a degraded roll with no challengers still uses the page, as a single text-only card with re-roll. Give every card the same anatomy, thesis, palette, materials, first viewport, honest risk, and the challengers' case lines (run the script with `--schema` for the exact shape); the page renders identity from these fields, and a challenger's catalog image rides as labeled inspiration, never as the promise of the build. Author `canonCard` too: the category standard as one honest card with the same anatomy; the page keeps it subordinate, and the counterweights still bind you. Run `node .claude/skills/impeccable/scripts/serve-question.mjs --start --payload <file>` (run it with `--schema` first for the exact payload shape). It daemonizes, prints the page URL and a key, and exits immediately; now open that URL for the user, in-app browser first, then the system opener, then showing the URL. Collect the choice with `--wait --key <key>`, repeating while it exits 3; the ANSWER prints as JSON. Exit 4 means the page was closed without an answer: re-present once through the structured question tool, and with no answer there either, proceed unattended with the assigned direction and state the assumptions. A harness that can leave a shell blocked in the background may instead run the script without `--start` and let it auto-open and block. Only a session where no browser can open at all, headless, CI, an eval worker, a remote shell with no display, puts the same decision through the structured question tool instead; the script self-detects these environments and exits 2 with that advice, so treat exit 2 as this fallback, never as an error to retry.
+The standing exit: every direction round offers one quiet, permanent alternative, the category standard, played straight. It is the user's door, never yours: never recommend it, never weigh it against the roll, never let it soften the dealt directions; the counterweights bind the unchosen default, not the chosen one. When the user takes it, in the canon action, a safer-steer, or plain words asking for the familiar or competitor-like path, convention becomes the commitment: ask once for two or three products this should sit alongside, make their craft level the bar, and execute the canon at full fidelity, without irony or smuggled quirk. A standing preference gets recorded as a brand commitment in PRODUCT.md. Re-roll eliminates every direction already shown, grounded and challenger alike; after two consecutive re-rolls, ask what quality is missing. You may re-roll on your own only on named factual grounds, when the assigned direction cannot carry the product's truth or task; taste is never grounds. The user may re-roll freely, and a user- or brief-pinned direction beats the roll, always. Present the decision visually: write an options payload with the assigned direction leading and its raised lines included, the pick card when one exists, the dealt challengers as alternates carrying their QUALITY BAR cards plus each challenger's verdict and kept line, re-roll with its safer and bolder registers, steer, plus canon enabled, and `followup: true` when the execution-contract round will follow (it does whenever image generation exists and no standing build-path preference is recorded); a degraded roll with no challengers still uses the page, as a single text-only card with re-roll. Give every card the same anatomy, thesis, palette, materials, first viewport, honest risk, and the challengers' case lines (run the script with `--schema` for the exact shape); the page renders identity from these fields, routes declined challengers to a demoted row on its own, and a challenger's catalog image rides as labeled inspiration, never as the promise of the build. Author `canonCard` too: the category standard as one honest card with the same anatomy; the page keeps it subordinate, and the counterweights still bind you. Run `node .claude/skills/impeccable/scripts/serve-question.mjs --start --payload <file>` (run it with `--schema` first for the exact payload shape). It daemonizes, prints the page URL and a key, and exits immediately; now open that URL for the user, in-app browser first, then the system opener, then showing the URL. Collect the choice with `--wait --key <key>`, repeating while it exits 3; the ANSWER prints as JSON. Exit 4 means the page was closed without an answer: re-present once through the structured question tool, and with no answer there either, proceed unattended with the assigned direction and state the assumptions. A harness that can leave a shell blocked in the background may instead run the script without `--start` and let it auto-open and block. Only a session where no browser can open at all, headless, CI, an eval worker, a remote shell with no display, puts the same decision through the structured question tool instead; the script self-detects these environments and exits 2 with that advice, so treat exit 2 as this fallback, never as an error to retry.
 
-When image generation exists, every card also declares a `sketch` path under `.impeccable/sketches/`, the canon card included. Where the harness sandboxes its shell, start the page through the least-sandboxed command path it offers: a sandboxed shell cannot bind the board's port, and the first-attempt failure costs a retry every session. Serve the page first, then produce the sketches; the page shimmer-waits per slot and the user may answer before they land. Render every sketch through one shared frame so the comparison stays about direction, never rendering luck: the requested surface's first viewport as a flat, matte design sketch in that card's own palette and type character, deliberately unfinished, no photorealism, no gloss, identical framing across cards; a candidate whose sketch looks more finished than the others has broken the comparison, not won it. The frame's aspect is the surface's own: a native app or mobile-first surface sketches portrait at its device viewport, a desktop web surface landscape, and the decision page adapts to either, so a phone screen sketched landscape is a broken frame, not a neutral default. The only legible text in a sketch is the product's real name and one real headline; every other text region is greeked, indistinct lines standing where copy will go, because a sketch that renders invented specs, prices, or dates puts claims in front of the user that PRODUCT.md never made. Produce in the order the user reads: the assigned card, then the hand, then canon, each file written the moment it is done. When the harness runs subagents in parallel, fan the set out as one agent per card: each spawn is the shipped asset producer with a single-sketch packet, that card's fields, PRODUCT.md, the shared frame, and the card's declared path, up to four in flight at once. A slot still empty when its agent returns is regenerated inline, and a slot still empty when the user answers is dropped without ceremony; no other supervision is owed. Without parallel subagents, generate in the main thread after serving, in the same reading order, and let the harness's own generation display carry the progress; the wait for the answer follows the last file. A sketch answers which world, never which composition: the comp round still renders its full set, and the chosen card's sketch seeds at most one probe. With no image generation, the cards carry their identity in palette chips and facts, and that page is complete, not a lesser version.
+When image generation exists, every card also declares a `sketch` path under `.impeccable/mocks/decision/` (the field keeps its wire name for compatibility; what it carries is the card's comp), the canon card included. Where the harness sandboxes its shell, start the page through the least-sandboxed command path it offers: a sandboxed shell cannot bind the board's port, and the first-attempt failure costs a retry every session. Serve the page first, then produce the comps; the page shimmer-waits per slot and the user may answer before they land. Each card's image is that direction's north-star comp at full fidelity, produced under the comp discipline in [visualize.md](visualize.md): the requested surface's first viewport, structure-led prompt, real product name and real content, no invented commercial claims, in that card's own palette, type character, and material world, committed all the way. Generation takes the same time at any fidelity, so an unfinished sketch pays sketch quality for comp cost; fairness between cards comes from equal fidelity in each card's own grammar, one surface, one aspect, never from shared unfinishedness. The frame's aspect is the surface's own: a native app or mobile-first surface comps portrait at its device viewport, a desktop web surface landscape, and the decision page adapts to either, so a phone screen comped landscape is a broken frame, not a neutral default. Produce in the order the user reads, the assigned card, then the pick, then the full-card hand, then canon, each file written with its prompt sidecar the moment it is done, so a re-roll's spend front-loads onto the cards read first; declined challengers get no comp, their catalog thumb is their face. When the harness runs subagents in parallel, fan the set out as one agent per card: each spawn is the shipped asset producer with a single-comp packet, that card's fields, PRODUCT.md, the shared frame, and the card's declared path, up to four in flight at once. A slot still empty when its agent returns is regenerated inline, and a slot still empty when the user answers is dropped without ceremony; no other supervision is owed. Without parallel subagents, generate in the main thread after serving, in the same reading order, and let the harness's own generation display carry the progress; the wait for the answer follows the last file. The chosen card's comp is not spent by the choice: on a comp-led build it enters the comp round as compositional option one, and on a code-led build it returns at the finish review as the critique reference, what the image dared that the build did not. The unchosen comps stay in `.impeccable/mocks/decision/` as the round's spent hand; they carry no approval and imply none. With no image generation, the cards carry their identity in palette chips and facts, and that page is complete, not a lesser version; the page then also demotes every challenger's catalog art to a labeled thumbnail on its own, because salience must encode the verdict, never the accident of which cards have images.
+
+The moment the direction lands, one more round on the same open table decides the execution contract. The direction payload declares `followup: true`, so the table stays open after the pick; deliver the build-path payload through `--update` immediately. Two text-only cards. **Comp-led**: a first-viewport comp is generated and it is law, the finish review audits the build against it; boldest composition on the table, fix rounds expected, motion at risk; choosing it makes the comp non-optional, no silent skipping. **Code-led**: no comp of this page and no apology for it; the QUALITY BAR boards still calibrate finish, and the ambition moves into the written contract, the FIRST VIEWPORT block plus a named signature interaction and motion grammar, which the finish reviewer audits in behavior; code-led is not a discount on commitment, the direction still lands fully committed in code. Lead with the chosen world's fit: a costume-heavy catalog world leads comp-led, a quiet or conventional direction leads code-led; the lead is a default, never a decision, and the user flips it freely. A standing preference, voiced once, is recorded as a brand commitment in PRODUCT.md and skips this round on later surfaces. Without image generation there is no fork and no round: code-led is the only path, stated in one line rather than asked. Only a detached table (`--start`) stays open for `--update`: a blocking serve or the structured-tool channel runs the build-path round as its own second question instead, and `followup: true` belongs only on a detached round.
 
 Catalog worlds are working systems, not mood references. When one survives, carry its palette and material, type and composition, topology, controls and state, and responsive rules into the product. When the source is itself an interface language, commit to its native grammar across navigation, content, controls, and states. Open the QUALITY BAR board and hero for the world you build the moment the choice lands, even if you viewed another card earlier; the ANSWER line names the chosen card's images (when the harness only reads files or runs sandboxed, download them into the workspace and open the relative path; sandboxed viewers reject absolute paths outside it). They set the craft level the build must reach, a rendered reference's finish, commitment, and art direction, never the composition; your surface serves this product.
 
@@ -80,7 +82,7 @@ If the work establishes durable strategy for a route or artifact, read its exist
 
 Keep the brief small: scope and visitor mode; audience, job, action/task, proof/content, and constraints; chosen direction and memorable moment; unresolved decisions. Do not copy global product truth or DESIGN.md tokens into it.
 
-Whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports, the locked direction is visualized before it is built, never skipped: load [visualize.md](visualize.md) and follow it, three compositional options rendered and put before the user for approval. This step is proven to produce the most compositional and ambitious work.
+On a comp-led build, whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports, the locked direction is visualized before it is built, never skipped: load [visualize.md](visualize.md) and follow it, three compositional options put before the user for approval, the chosen card's decision comp plus two variations. This step is proven to produce the most compositional and ambitious work. On a code-led build the comp round is skipped by contract, never by drift: the ambition it would have carried lives in the direction contract's FIRST VIEWPORT block and named signature interaction, and the finish reviewer audits those promises in behavior.
 
 For `shape`, return the selected direction to [shape.md](shape.md) and stop before persistence or implementation.
 
@@ -103,8 +105,8 @@ Preserve semantics, accessibility, performance, responsiveness, project conventi
 
 ## 7. Inspect and finish
 
-Inspect desktop and mobile in one batched screenshot round, critique the render against the user's request and the direction contract, fix material gaps, and confirm with one final round; two rounds is the ceiling, and fixes batch between them rather than earning per-tweak screenshots. When an approved comp exists, the critique is a side-by-side: view the comp region and the build region together, the hero and each section as its own crop at legible scale, never one full-page thumbnail, which hides exactly the failures that matter, crude controls, wrong lettering character, flattened material, behind a superficially similar section order. On a Persuade surface, verify the mode did its job: a first-time visitor should know what this is, why it matters, and what to do within seconds, in the form's own vocabulary.
+Inspect the surface's target sizes in one batched screenshot round: desktop and mobile on the web; on a native platform (`ios` / `android` / `adaptive`), the shipped device classes per OS, captured from the simulator or emulator the way the platform reference's Verifying the build section describes. Critique the render against the user's request and the direction contract, fix material gaps, and confirm with one final round; two rounds is the ceiling, and fixes batch between them rather than earning per-tweak screenshots. When an approved comp exists, the critique is a side-by-side: view the comp region and the build region together, the hero and each section as its own crop at legible scale, never one full-page thumbnail, which hides exactly the failures that matter, crude controls, wrong lettering character, flattened material, behind a superficially similar section order. On a Persuade surface, verify the mode did its job: a first-time visitor should know what this is, why it matters, and what to do within seconds, in the form's own vocabulary.
 
-After the second inspection round the build thread's polishing is over: no further defect hunts, micro-edit scripts, or rebuilds here; whatever remains ships through the handoffs, where a fresh context does the finding better and cheaper. Where this harness runs no design hook, run `node .claude/skills/impeccable/scripts/detect.mjs --json` on the changed targets once here, fix what is mechanical, and pass the remaining findings to the reviewer; a hookless build that skips this ships every tell the hook exists to catch. Capture desktop and mobile screenshots to files, then spawn the shipped finish reviewer, `impeccable-finish-reviewer` (`impeccable_finish_reviewer` in codex; `/impeccable-finish-reviewer` in Cursor; on GitHub Copilot say "Use the impeccable-finish-reviewer agent"), with the original request, confirmed answers, the artifact path, the screenshot paths, its direction contract, existing hook findings, the QUALITY BAR card and approved comp paths, and the craft-floor reference path. The reviewer has no browser; screenshots you fail to pass are checks it cannot run. Never read the shipped agents' definition files before spawning; the harness loads them at spawn, and you owe only the input packet. Wait on any agent with one long timeout rather than a loop of short polls, and spend the wait on the next independent step. Verify its return carries the five contract sections; on an empty or thrashed return, respawn once with the same inputs before doing anything else. This review never runs inside the build thread and never inherits it: spawn the reviewer fresh, with no forked conversation history (`fork_turns: 0` in codex); a reviewer that inherits your transcript inherits your framing, your optimism, and your abstractions, and everything it needs travels in the inputs above. Only a harness whose tool surface has no subagent capability at all substitutes a fresh in-thread pass after stepping fully out of the build context, run from [degraded/finish-reviewer.md](degraded/finish-reviewer.md), and a substituted or failed-and-replaced review is disclosed in one line at finish, never silently. When the reviewer's first material fix is a rebuild directive, fidelity failed wholesale rather than in patches, so skip the fix batch and execute the rebuild immediately: re-derive the named regions, produce the named assets, and send the result back for a verdict, telling the user what is happening rather than asking permission to fix a failure. The user is consulted only when a second rebuild directive arrives, both verdicts on the table, or when rebuilding would discard content the user approved. Otherwise apply the material fixes in one batch, rebuild once, and recapture the same viewports. A recapture measures positions, loading, and overflow; it cannot measure whether a fix reached the quality the finding named, so send the recaptured screenshots back to the same reviewer for a verdict scoring every material fix resolved, partial, or unresolved (through the harness's agent continuation; without one, run the scoring fresh from [degraded/finish-reviewer.md](degraded/finish-reviewer.md)'s Verdict Pass). Fixes scored partial or unresolved get another batch, recapture, and verdict. Two rounds is the budget an unattended run ends at; an attended session's ceiling belongs to the user, so when the second verdict still lists open items, put the table in front of them and let them choose between shipping as it stands and funding another round. Whoever is deciding, stop the moment a round resolves nothing, and the reviewer's findings are the only list you work from, never your own re-opened hunt. Report the final verdict table to the user as it stands, open items included, under the reviewer's own disposition word: a table with open material findings is never announced as a pass, and never under a softer label than the reviewer wrote. Do not run a second detector.
+After the second inspection round the build thread's polishing is over: no further defect hunts, micro-edit scripts, or rebuilds here; whatever remains ships through the handoffs, where a fresh context does the finding better and cheaper. On the web, where this harness runs no design hook, run `node .claude/skills/impeccable/scripts/detect.mjs --json` on the changed targets once here, fix what is mechanical, and pass the remaining findings to the reviewer; a hookless web build that skips this ships every tell the hook exists to catch. A native platform skips the detector entirely: it reads HTML and CSS and has no verdict on native code, so the reviewer's floor check is the only slop gate and the input packet says so. Capture the screenshots into `.impeccable/review/`, one file per captured viewport (on the web, `desktop.png` and `mobile.png`; on native, one per device class, such as `phone.png` and `tablet.png`, suffixed per OS on adaptive), creating that directory when the harness does not; the paths you pass the reviewer are its spec, and that directory is where it looks when a passed path is missing. Then spawn the shipped finish reviewer, `impeccable-finish-reviewer` (`impeccable_finish_reviewer` in codex; `/impeccable-finish-reviewer` in Cursor; on GitHub Copilot say "Use the impeccable-finish-reviewer agent"), with the original request, confirmed answers, the artifact path, the screenshot paths, its direction contract, existing hook findings, the QUALITY BAR card and approved comp paths (on a code-led build there is no approved comp; the chosen decision comp rides in that slot as the critique reference, named as such), the craft-floor reference path, and on a native platform the platform reference path(s), [ios.md](ios.md) / [android.md](android.md), both on adaptive, plus one line saying no detector ran, so the reviewer judges in the platform's conventions rather than the web's. The reviewer has no browser; screenshots you fail to pass are checks it cannot run. Never read the shipped agents' definition files before spawning; the harness loads them at spawn, and you owe only the input packet. Wait on any agent with one long timeout rather than a loop of short polls, and spend the wait on the next independent step. Verify its return carries the five contract sections; on an empty or thrashed return, respawn once with the same inputs before doing anything else. This review never runs inside the build thread and never inherits it: spawn the reviewer fresh, with no forked conversation history (`fork_turns: 0` in codex); a reviewer that inherits your transcript inherits your framing, your optimism, and your abstractions, and everything it needs travels in the inputs above. Only a harness whose tool surface has no subagent capability at all substitutes a fresh in-thread pass after stepping fully out of the build context, run from [degraded/finish-reviewer.md](degraded/finish-reviewer.md), and a substituted or failed-and-replaced review is disclosed in one line at finish, never silently. When the reviewer's first material fix is a rebuild directive, fidelity failed wholesale rather than in patches, so skip the fix batch and execute the rebuild immediately: re-derive the named regions, produce the named assets, and send the result back for a verdict, telling the user what is happening rather than asking permission to fix a failure. The user is consulted only when a second rebuild directive arrives, both verdicts on the table, or when rebuilding would discard content the user approved. Otherwise apply the material fixes in one batch, rebuild once, and recapture the same viewports over the same files. A recapture measures positions, loading, and overflow; it cannot measure whether a fix reached the quality the finding named, so send the recaptured screenshots back to the same reviewer for a verdict scoring every material fix resolved, partial, or unresolved (through the harness's agent continuation; without one, run the scoring fresh from [degraded/finish-reviewer.md](degraded/finish-reviewer.md)'s Verdict Pass). Fixes scored partial or unresolved get another batch, recapture, and verdict. Two rounds is the budget an unattended run ends at; an attended session's ceiling belongs to the user, so when the second verdict still lists open items, put the table in front of them and let them choose between shipping as it stands and funding another round. Whoever is deciding, stop the moment a round resolves nothing, and the reviewer's findings are the only list you work from, never your own re-opened hunt. Report the final verdict table to the user as it stands, open items included, under the reviewer's own disposition word: a table with open material findings is never announced as a pass, and never under a softer label than the reviewer wrote. Do not run a second detector.
 
 Then spawn the shipped documenter, `impeccable-documenter` (`impeccable_documenter` in codex), with the project root, the artifact path, the direction contract, PRODUCT.md, the [document.md](document.md) reference path, and the boundary to write at; it records DESIGN.md and the sidecar from the built world, ground truth over intention; without subagents the pass runs from [degraded/documenter.md](degraded/documenter.md). A clean detector pass is not finished; finished is the contract kept, the comp honored, the review closed, and the system recorded.

.claude/skills/impeccable/reference/polish.md

@@ -19,7 +19,7 @@ Fix the cause at the narrowest correct level. Ask when a binding system principl
 
 ## 2. Gather the evidence
 
-Use the feature yourself at representative desktop and mobile sizes. Determine:
+Use the feature yourself at the surface's representative sizes: desktop and mobile on the web; on a native platform (`ios` / `android` / `adaptive`), the shipped device classes on the simulator, emulator, or hardware, captured per the platform reference's Verifying the build section. Determine:
 
 - whether the path is functionally complete;
 - the intended quality bar and time available;
@@ -86,10 +86,10 @@ Do not perfect one corner while leaving the rest below the same quality bar.
 
 Walk the complete path again with mouse, keyboard, and touch where applicable. Check:
 
-- mobile, intermediate, and wide layouts;
+- mobile, intermediate, and wide layouts on the web; phone and tablet size classes in both supported orientations on native;
 - loading, empty, error, success, disabled, long-content, and missing-content states;
 - zoom, contrast, focus, semantics, and screen-reader names;
-- console errors, layout shift, interaction latency, image loading, and supported browsers;
+- console errors, layout shift, interaction latency, and image loading everywhere; supported browsers on the web; supported OS versions, runtime warnings, and dropped frames on native;
 - agreement with DESIGN.md, neighboring features, and the user's scope.
 
 Follow the quality guidance supplied by `context.mjs` and hooks, then run any other relevant QA commands. Context requests a manual scan only when no automatic detector is active; never add another detector pass. Fix real defects and document only narrow intentional exceptions. A clean scan does not replace visual judgment.

.claude/skills/impeccable/reference/visualize.md

@@ -1,12 +1,12 @@
 # Visualize: Direction Comps & Asset Production
 
-Load this from [new-work.md](new-work.md) whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports. PRODUCT.md and DESIGN.md are preconditions. New-work has already resolved the visual world; this file must not reopen it.
+Load this from [new-work.md](new-work.md) on a comp-led build, when image generation is available (a harness-native tool or the API fallback context.mjs reports). A code-led execution contract skips this file by design, not by drift: its ambition lives in the written direction contract and is audited in behavior, so do not load it for a code-led round. PRODUCT.md and DESIGN.md are preconditions. New-work has already resolved the visual world; this file must not reopen it.
 
 The purpose of a probe is to test composition, narrative, hierarchy, density, focal moment, signature use, and image requirements. It is not a second identity workshop. Keep DESIGN.md's palette, typography direction, material language, component character, imagery stance, and motion grammar fixed.
 
 ## Generate three compositional options
 
-Render three distinct high-fidelity north-star comps of the requested surface, with whatever generation capability exists, saved under `.impeccable/mocks/` so they survive the session. Comp at the surface's own viewport: portrait at device size for a native app or mobile-first surface, desktop landscape otherwise; a phone screen comped landscape misstates the composition before anything gets built against it. Comps are the build thread's own work, never delegated: the thread that writes the comp prompts holds the direction's full context, and it has already seen every comp when the build starts. Open every image you produce or reference by its workspace-relative path, never an absolute one: sandboxed viewers reject absolute paths, and everything under the project root has a relative path. Base them on the real content and the surface concepts already developed with the user. Three is the number: one comp invites rubber-stamping, and the spread between three is what surfaces the composition worth building. A decision-page sketch is not a probe: it chose the direction at deliberately unfinished fidelity, so the three comps render regardless, and the chosen card's sketch seeds at most one of them.
+Render three distinct high-fidelity north-star comps of the requested surface, with whatever generation capability exists, saved under `.impeccable/mocks/` so they survive the session. Comp at the surface's own viewport: portrait at device size for a native app or mobile-first surface, desktop landscape otherwise; a phone screen comped landscape misstates the composition before anything gets built against it. Comps are the build thread's own work, never delegated: the thread that writes the comp prompts holds the direction's full context, and it has already seen every comp when the build starts. Open every image you produce or reference by its workspace-relative path, never an absolute one: sandboxed viewers reject absolute paths, and everything under the project root has a relative path. Base them on the real content and the surface concepts already developed with the user. Three is the number: one comp invites rubber-stamping, and the spread between three is what surfaces the composition worth building. The chosen card's decision comp is the first of the three: it already renders this direction at full fidelity under this file's discipline, so this round generates two more that vary what the first held fixed, and all three go to the approval point together. Only a round that arrives with no decision comp, a degraded roll, an identity-mode page, a direction pinned without the decision round, renders all three here.
 
 - A comp is a designed surface, not a picture of the subject. Lead the generation prompt with the surface's own structure, whatever regions this design actually has, named in order with their scale relationships; a page with no navigation states that instead of inventing one, and an unconventional surface states its unconventional skeleton. A prompt that leads with the world's atmosphere gets a vignette back: the model paints the fish market instead of the fish market's website. Self-check every render: if it could hang as a poster, or reads as a photograph or scene with some text on it, it is not a comp; regenerate with the layout scaffold stated more literally.
 - When the user shortlisted multiple concepts, spread the three across them.
@@ -22,7 +22,7 @@ Show the three together: in the harness when it can display images, otherwise on
 
 Do not begin code until the user approves a direction or explicitly delegates the choice. If they delegate, choose using the task brief, PRODUCT.md, and DESIGN.md, and state the evidence. Approval refines the task concept; it does not modify DESIGN.md.
 
-This approval point has no substitute and no skip condition. When the structured question tool errors, fall back to the decision page; only after both fail may you treat the choice as delegated, and a delegated pick is still recorded exactly as an approval is and disclosed in your first reply, not your last. The finish reviewer treats a build with generated comps and no recorded approval as carrying a material finding.
+This approval point has no substitute and no skip condition. When the structured question tool errors, fall back to the decision page; only after both fail may you treat the choice as delegated, and a delegated pick is still recorded exactly as an approval is and disclosed in your first reply, not your last. The finish reviewer treats a build whose comp round produced comps with no recorded approval as carrying a material finding; decision comps under `.impeccable/mocks/decision/` are the direction round's hand, not comp-round output, and imply no approval on their own.
 
 After approval, record the choice where tools can find it: the approved comp's path goes in the surface brief, and the approved comp's `.json` prompt sidecar gains `"approved": true` (every comp generated through `generate-image.mjs` has one; create it if a native tool didn't). The sidecar travels with the mocks folder, so the approval survives sessions and machines that never see the brief. Then summarize the composition and the parts of the comp that must not be literalized, return to new-work.md, record the direction contract from the approved surface concept, and build.
 

.claude/skills/impeccable/scripts/concept-seed.mjs

@@ -31,6 +31,16 @@
  *     recomputes what rounds 0..n-1 drew, excludes all of it, and rolls a
  *     fresh assigned index, challengers, and compositions. One base key therefore
  *     reproduces the entire chain of rounds.
+ *   - REGISTER (--register safer|bolder): the user's steering on the
+ *     familiar-to-bold axis, applied to a re-roll round. A register changes
+ *     only what this round instructs, never what it dealt: the same key and
+ *     reroll count reproduce the same deal whatever the register, so the
+ *     exclusion chain never forks. bolder presents the dealt foreign forms
+ *     as the whole hand (first-dealt leads, dice-assigned by deal order);
+ *     safer spends the dealt hand unseen and presents the familiar register,
+ *     the model's conventional grounded candidates plus the canon against
+ *     named competitors, the one sanctioned lineup of the model's own list.
+ *     Registers are user-requested, never pre-selected by the model.
  *   - RATINGS: the reviewer's approval ratings weight the challenger draw
  *     (3-star doubles the odds, 1-star sits out); the approved pool itself
  *     is unchanged.
@@ -41,7 +51,9 @@
  *   node scripts/concept-seed.mjs --scope surface --mode operate --grain flow
  *   node scripts/concept-seed.mjs --scope direction --candidate-count 6
  *   node scripts/concept-seed.mjs --scope direction --mode persuade --from <key> --reroll 1
- *   node scripts/concept-seed.mjs --chosen <challenger-id> --from <key> --scope direction
+ *   node scripts/concept-seed.mjs --scope direction --mode persuade --from <key> --reroll 1 --register bolder
+ *   node scripts/concept-seed.mjs --chosen <challenger-id> --kind challenger --from <key> --scope direction
+ *   node scripts/concept-seed.mjs --kind assigned --from <key> --scope direction
  *
  * --grain names how much of the product is in play: product, flow, view, or
  * region. A docs site, an onboarding flow, a landing page and a data table are
@@ -62,8 +74,13 @@
  * Challenger data resolves in order: a local catalog directory (the private
  * service repo, evals, and tests set IMPECCABLE_CATALOG_DIR), then the roll
  * API at impeccable.style, then a degraded assignment-only seed when both are
- * unavailable. --chosen sends the anonymous choice ping for API-dealt rolls;
- * DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY disables it.
+ * unavailable. The anonymous choice ping fires once per resolved attended
+ * round on API-dealt rolls: --kind names which card class won (assigned,
+ * pick, challenger, canon) so share metrics have a denominator, --chosen
+ * carries the catalog id when a dealt challenger won, and --register rides
+ * along when the round came from a steered hand. Grounded candidates' names
+ * never leave the machine. DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY disables
+ * the ping entirely.
  *
  * Env vars:
  *   IMPECCABLE_CONCEPT_SEED — same as --from; for reproducible eval runs.
@@ -172,17 +189,35 @@ function telemetryDisabled() {
   return Boolean(process.env.IMPECCABLE_NO_TELEMETRY || process.env.DO_NOT_TRACK);
 }
 
-// Anonymous choice ping: records only that a dealt world was selected.
+// Anonymous choice ping: one per resolved attended direction round. kind
+// says which card class won (assigned / pick / challenger / canon), so
+// pick-share and canon-share have a denominator; chosenId rides along only
+// when a dealt catalog world won, and register only when the round came from
+// a steered hand. Grounded candidates' names never leave the machine: they
+// are derived from the user's project, so the ping carries the kind alone.
 // Fire-and-forget; never fails the caller.
-export async function pingChosen({ chosenId, key, scope, mode }) {
-  if (telemetryDisabled() || !chosenId) return false;
+const PING_KINDS = new Set(['assigned', 'pick', 'challenger', 'canon']);
+export async function pingChosen({ chosenId, key, scope, mode, kind, register }) {
+  if (telemetryDisabled()) return false;
+  if (kind && !PING_KINDS.has(kind)) return false;
+  if (register && register !== 'safer' && register !== 'bolder') return false;
+  // Legacy shape: a bare challenger id with no kind stays a valid ping.
+  if (!chosenId && !kind) return false;
+  if ((kind === 'challenger' || !kind) && !chosenId) return false;
   const controller = new AbortController();
   const timer = setTimeout(() => controller.abort(), apiBudgetMs());
   try {
     await fetch(`${API_BASE}/chosen`, {
       method: 'POST',
       headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify({ chosenId, key, scope, mode }),
+      body: JSON.stringify({
+        ...(chosenId ? { chosenId } : {}),
+        key,
+        scope,
+        mode,
+        ...(kind ? { kind } : {}),
+        ...(register ? { register } : {}),
+      }),
       signal: controller.signal,
     });
     return true;
@@ -260,6 +295,7 @@ export function renderConceptSeed({
   scope = 'surface',
   key = process.env.IMPECCABLE_CONCEPT_SEED || crypto.randomBytes(4).toString('hex'),
   reroll = 0,
+  register = null,
   mode = null,
   grain = null,
   platform = null,
@@ -273,6 +309,15 @@ export function renderConceptSeed({
   if (!Number.isInteger(reroll) || reroll < 0) {
     throw new Error('concept-seed: --reroll must be a non-negative integer');
   }
+  if (register !== null && register !== 'safer' && register !== 'bolder') {
+    throw new Error('concept-seed: --register must be safer or bolder');
+  }
+  if (register !== null && reroll < 1) {
+    throw new Error('concept-seed: --register steers a re-roll round; pass --reroll <n> with it');
+  }
+  if (register !== null && scope !== 'direction') {
+    throw new Error('concept-seed: --register applies to direction rounds only');
+  }
   if (mode !== null && !SEED_MODES.has(mode)) {
     throw new Error('concept-seed: --mode must be persuade, operate, read, or experience');
   }
@@ -326,6 +371,7 @@ export function renderConceptSeed({
         scope,
         key,
         reroll,
+        register,
         mode,
         grain,
         platform,
@@ -357,7 +403,11 @@ export function renderConceptSeed({
   survive the current task plus navigation, quiet and dense content,
   interaction and state, and a substantially different future surface. In an
   attended run, present the assigned direction fully committed and offer
-  re-roll; never present a ranked lineup to choose from. Re-roll yourself only
+  re-roll. You may add ONE card for your top-ranked grounded candidate when
+  it is not the assigned direction, kicker MY PICK, with an honest risk line
+  naming its familiarity; one pick card, never a ranked lineup, and the pick
+  never takes the lead position. When the assignment IS your top candidate,
+  there is no pick card. Re-roll yourself only
   on named factual grounds, when the assignment cannot carry the product's
   truth or task; taste is never grounds.`
   : `After ordering the task's grounded structural candidates by resonance,
@@ -374,7 +424,16 @@ export function renderConceptSeed({
   conflicts. Weigh the fused result against the assigned direction on exactly
   two axes, audience identification and product clarity. Losing to strong
   grounded material is a valid outcome; beating a thin or tool-monoculture
-  list is the point. A fused challenger that wins both axes becomes the build.`
+  list is the point. A fused challenger that wins both axes becomes the build.
+  Close the weighing with a verdict per challenger, decided before any
+  borrowing is considered: wins (beats the assigned direction on both axes),
+  competitive (holds one axis), or declined (loses both). A declined
+  challenger is not spent: name the one discipline of its system the assigned
+  direction lacks, and raise the assigned direction to match before
+  presenting it. A donation transfers ambition and system discipline, never
+  the challenger's clothes; one world owns the page. Write each raise as its
+  own named line on the presented direction, and carry every verdict, kept
+  line, and raise into the decision page payload.`
   : `A challenger wins only when its fused result beats the grounded list on
   audience identification and product clarity. It may change task topology or
   interaction, but never the committed visual identity.`;
@@ -399,8 +458,39 @@ Ambitious motion, spatial media, or interaction is welcome when it strengthens
 the product without weakening semantics, performance, or fallback behavior.`;
 
   if (!data) {
-    return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: degraded; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''} --candidate-count ${candidateCount})
-ASSIGNED INDEX: ${buildIndex}
+    // A degraded roll can still serve the safer register, which needs no
+    // catalog at all: the assignment machinery is suppressed entirely, the
+    // same as the non-degraded safer round, because emitting both "the user
+    // picks" and a mandatory numbered build order hands the model two
+    // contradicting instructions and the mandatory one tends to win. The
+    // bolder register is exactly the thing degradation took away, so it
+    // falls back to a plain grounded round, disclosed.
+    const degradedHeader = `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: degraded; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''}${register ? ` --register ${register}` : ''} --candidate-count ${candidateCount})`;
+    if (register === 'safer') {
+      return `${degradedHeader}
+SAFER REGISTER (user-requested): the assigned index is suspended this
+  round; the user picks, and no candidate is mandated. Present the familiar
+  register: your remaining grounded candidates from the conventional end, at
+  most three, as full cards with an honest risk line each, plus the canon
+  executed against two or three named competitors. This is the one sanctioned
+  lineup of your own ranked candidates; it exists only by this explicit
+  request. When the user voices a standing preference for it, record a brand
+  commitment in PRODUCT.md.
+${authorityInstruction}
+A user- or brief-pinned decision beats the roll, always.
+REGISTER (restated for truncated readers): safer, user-requested; the
+assigned index is suspended this round and the user picks; seed key ${key}.
+`;
+    }
+    const degradedRegister = register === 'bolder'
+      ? `BOLDER REGISTER UNAVAILABLE: bolder deals foreign forms, and this roll ran
+  degraded with no catalog and no roll service, so there is nothing bold to
+  deal. Tell the user, then run this round as a plain grounded re-roll; the
+  assignment below applies.
+`
+      : '';
+    return `${degradedHeader}
+${degradedRegister}ASSIGNED INDEX: ${buildIndex}
   ${promotedInstruction}
   The assignment exists to refuse the model's ranking rut, never to outrank
   the user or the brief. Never expose assignment metadata in user-facing labels.
@@ -471,34 +561,76 @@ structure only, never a palette, typeface, or material. Treat them as serious
 rivals to your habitual layout, and keep only what makes this product clearer.${grainNote}\n`
     : '';
   const rerollBlock = reroll > 0
-    ? `RE-ROLL ROUND ${reroll}: every candidate presented in earlier rounds, grounded
-  and challenger alike, is eliminated and may not return reworded. Derive
+    ? `RE-ROLL ROUND ${reroll}${register ? ` (${register.toUpperCase()} REGISTER, user-requested)` : ''}: every candidate presented in earlier rounds, grounded
+  and challenger alike, is eliminated and may not return reworded.${register ? '' : ` Derive
   genuinely new grounded candidates from unexplored angles before judging
-  these fresh challengers.\n`
+  these fresh challengers.`}\n`
     : '';
+  // A register swaps the round's presentation, never its deal: the assigned
+  // index and challenger fetch stay identical so the chain reproduces, and
+  // only the instructions change.
+  const saferBlock = `SAFER REGISTER: the user asked for the familiar end of the spectrum, so this
+  round's dealt hand is spent unseen, stays excluded from future rounds, and
+  is not printed. The assigned index is suspended this round; the user picks. Present the familiar register: your remaining grounded
+  candidates from the conventional end, at most three, as full cards with an
+  honest risk line each, plus the canon executed against two or three named
+  competitors. This is the one sanctioned lineup of your own ranked
+  candidates; it exists only by this explicit request. When the user voices a
+  standing preference for it, record a brand commitment in PRODUCT.md.`;
+  const bolderBlock = `BOLDER REGISTER: the user asked for foreign forms at full commitment, so no
+  grounded direction is presented this round and the assigned index is
+  suspended. The hand is every dealt challenger below, each fused with the
+  product and presented as a full card; the FIRST dealt challenger leads, an
+  assignment by deal order, so the dice still choose. Verdicts and donations
+  apply between the challengers, weighed against the leader. The pick card
+  sits out; the canon stays, as always.`;
   const telemetryBlock = data.source === 'api'
-    ? `TELEMETRY: if the resolved direction uses one of these challengers, rerun
-  this script once with --chosen <challenger-id> --from ${key} --scope ${scope}${mode ? ` --mode ${mode}` : ''}
-  after resolution. The ping is anonymous (chosen id only) and is skipped
-  automatically when DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY is set.\n`
+    ? `TELEMETRY: after the user's choice resolves, rerun this script once with
+  --kind <assigned|pick|challenger|canon> --from ${key} --scope ${scope}${mode ? ` --mode ${mode}` : ''},
+  adding --chosen <challenger-id> when a dealt challenger won and keeping
+  --register <safer|bolder> when the resolved round came from a steered hand.
+  One ping per resolved attended round. The ping is anonymous, the card kind
+  plus the catalog id when one won; your grounded candidates' names never
+  leave the machine, and the ping is skipped automatically when DO_NOT_TRACK
+  or IMPECCABLE_NO_TELEMETRY is set.\n`
     : '';
-  return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: ${data.source}; approved pool: ${data.poolRevision}; ${data.approvedCount}/${data.catalogCount} human-approved; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''} --candidate-count ${candidateCount} to reproduce this roll against this catalog revision)
-${rerollBlock}ASSIGNED INDEX: ${buildIndex}
+  const assignedBlock = register === null
+    ? `ASSIGNED INDEX: ${buildIndex}
   ${promotedInstruction}
   The assignment exists to refuse the model's ranking rut, never to outrank
-  the user or the brief. Never expose assignment metadata in user-facing labels.
-CHALLENGERS:
+  the user or the brief. Never expose assignment metadata in user-facing labels.`
+    : register === 'safer' ? saferBlock : bolderBlock;
+  // A bolder round has no assigned grounded direction, so the generic
+  // weighing instruction (which measures against the assignment) would
+  // contradict the register; the bolder variant weighs against the leader.
+  const bolderChallengerInstruction = `Fuse each challenger before judging it: the challenger supplies the form
+  and its system grammar, the product supplies every fact, and clarity wins
+  conflicts. Weigh every fused challenger against the fused LEADER, the first
+  dealt, on exactly two axes, audience identification and product clarity;
+  verdicts and donations apply between the challengers, and one that beats
+  the leader on both axes presents as the hand's strongest alternate.`;
+  const roundChallengerInstruction = register === 'bolder' ? bolderChallengerInstruction : challengerInstruction;
+  const challengerSection = register === 'safer'
+    ? ''
+    : `CHALLENGERS:
 ${data.challengers.map(renderChallenger).join('\n')}
-${compositionBlock}${challengerInstruction}
+${compositionBlock}${roundChallengerInstruction}
 When you can view images, open the QUALITY BAR board and hero for any
 challenger you weigh seriously and for the world you build. They exist as a
 craft bar, the finish level and commitment the build is expected to reach,
 never as a mockup to copy; your surface serves this product, not that render.
-${authorityInstruction}
+`;
+  const restated = register === null
+    ? `ASSIGNED INDEX (restated for truncated readers): ${buildIndex}. Build candidate
+${buildIndex} of your own grounded list; seed key ${key}.`
+    : `REGISTER (restated for truncated readers): ${register}, user-requested; the
+assigned index is suspended this round; seed key ${key}.`;
+  return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: ${data.source}; approved pool: ${data.poolRevision}; ${data.approvedCount}/${data.catalogCount} human-approved; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''}${register ? ` --register ${register}` : ''} --candidate-count ${candidateCount} to reproduce this roll against this catalog revision)
+${rerollBlock}${assignedBlock}
+${challengerSection}${authorityInstruction}
 ${richnessInstruction}
 ${telemetryBlock}A user- or brief-pinned decision beats the roll, always.
-ASSIGNED INDEX (restated for truncated readers): ${buildIndex}. Build candidate
-${buildIndex} of your own grounded list; seed key ${key}.
+${restated}
 `;
 }
 
@@ -507,19 +639,25 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
   const fromIdx = args.indexOf('--from');
   const scopeIdx = args.indexOf('--scope');
   const rerollIdx = args.indexOf('--reroll');
+  const registerIdx = args.indexOf('--register');
   const modeIdx = args.indexOf('--mode');
   const grainIdx = args.indexOf('--grain');
   const platformIdx = args.indexOf('--platform');
   const candidateCountIdx = args.indexOf('--candidate-count');
   const chosenIdx = args.indexOf('--chosen');
+  const kindIdx = args.indexOf('--kind');
   try {
-    if (chosenIdx !== -1) {
+    if (chosenIdx !== -1 || kindIdx !== -1) {
       // Choice ping: always exits 0, telemetry must never fail a design flow.
+      // --kind alone pings a non-challenger outcome (assigned/pick/canon);
+      // --chosen alone stays the legacy challenger-win ping.
       const sent = await pingChosen({
-        chosenId: args[chosenIdx + 1],
+        chosenId: chosenIdx !== -1 ? args[chosenIdx + 1] : undefined,
         key: fromIdx !== -1 ? args[fromIdx + 1] : undefined,
         scope: scopeIdx !== -1 ? args[scopeIdx + 1] : undefined,
         mode: modeIdx !== -1 ? args[modeIdx + 1] : undefined,
+        kind: kindIdx !== -1 ? args[kindIdx + 1] : undefined,
+        register: registerIdx !== -1 ? args[registerIdx + 1] : undefined,
       });
       process.stdout.write(sent ? 'choice recorded\n' : 'choice ping skipped\n');
     } else {
@@ -542,6 +680,7 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
           ? args[fromIdx + 1]
           : (process.env.IMPECCABLE_CONCEPT_SEED || crypto.randomBytes(4).toString('hex')),
         reroll: rerollIdx !== -1 ? Number(args[rerollIdx + 1]) : 0,
+        register: registerIdx !== -1 ? args[registerIdx + 1] : null,
         mode: modeIdx !== -1 ? args[modeIdx + 1] : null,
         grain: grainIdx !== -1 ? args[grainIdx + 1] : null,
         platform: platformIdx !== -1 ? args[platformIdx + 1] : null,
@@ -553,6 +692,13 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
     process.exitCode = 1;
   }
   // A raced-out fetch may still hold a socket; exit explicitly so the CLI
-  // never lingers on a dead network path after output is written.
+  // never lingers on a dead network path after output is written. Destroy
+  // fetch's global undici dispatcher first: process.exit() with a live
+  // keep-alive socket trips a libuv assertion on Windows and aborts the
+  // process after a successful roll (nodejs/node#56645).
+  const dispatcher = globalThis[Symbol.for('undici.globalDispatcher.1')];
+  if (dispatcher && typeof dispatcher.destroy === 'function') {
+    try { await dispatcher.destroy(); } catch { /* exit regardless */ }
+  }
   process.exit(process.exitCode ?? 0);
 }

.claude/skills/impeccable/scripts/context-signals.mjs

@@ -22,7 +22,7 @@ import path from 'node:path';
 import { fileURLToPath } from 'node:url';
 import { execFileSync } from 'node:child_process';
 import { loadContext, extractPlatform } from './context.mjs';
-import { getCritiqueDir } from './lib/impeccable-paths.mjs';
+import { readLatestSnapshotAcrossTargets } from './critique-storage.mjs';
 
 /** Is there code here at all, or just context files / an empty repo? */
 function hasCode(cwd) {
@@ -34,23 +34,13 @@ function hasCode(cwd) {
 }
 
 /**
- * The most recent critique snapshot across all targets. Filenames are
- * timestamp-prefixed (`<iso>__<slug>.md`), so a lexical sort is chronological.
- * Parses the small frontmatter for score + P0/P1 counts.
+ * Summarize the most recent critique snapshot across all targets.
  */
 function latestCritique(cwd) {
   try {
-    const dir = getCritiqueDir(cwd);
-    if (!fs.existsSync(dir)) return null;
-    const files = fs.readdirSync(dir).filter((f) => f.endsWith('.md')).sort();
-    if (!files.length) return null;
-    const newest = files[files.length - 1];
-    const text = fs.readFileSync(path.join(dir, newest), 'utf-8');
-    const front = text.split('---')[1] || '';
-    const get = (k) => {
-      const m = front.match(new RegExp(`^${k}:\\s*(.+)$`, 'm'));
-      return m ? m[1].trim() : null;
-    };
+    const latest = readLatestSnapshotAcrossTargets({ cwd });
+    if (!latest) return null;
+    const get = (key) => latest.meta[key] ?? null;
     const num = (v) => {
       const n = Number(v);
       return Number.isFinite(n) ? n : null;
@@ -61,7 +51,7 @@ function latestCritique(cwd) {
       p0: num(get('p0')),
       p1: num(get('p1')),
       timestamp: get('timestamp'),
-      file: path.relative(cwd, path.join(dir, newest)),
+      file: path.relative(cwd, latest.path),
     };
   } catch {
     return null;

.claude/skills/impeccable/scripts/critique-storage.mjs

@@ -105,28 +105,37 @@ function parseFrontmatter(text) {
 }
 
 /**
- * Return all snapshot files for `slug`, sorted oldest → newest.
+ * Return snapshot files matching `suffix`, sorted oldest → newest.
  */
-function listSnapshotsForSlug(slug, cwd) {
+const SNAPSHOT_FILENAME = /^\d{4}-\d{2}-\d{2}T\d{2}-\d{2}-\d{2}Z__.+\.md$/;
+
+function listSnapshots(suffix, cwd) {
   const dir = getCritiqueDir(cwd);
   if (!fs.existsSync(dir)) return [];
-  const suffix = `__${slug}.md`;
   return fs.readdirSync(dir)
-    .filter((f) => f.endsWith(suffix))
+    .filter((f) => SNAPSHOT_FILENAME.test(f) && f.endsWith(suffix))
     .sort()
     .map((f) => path.join(dir, f));
 }
 
+function readLatestSnapshotMatching(suffix, cwd) {
+  const filePath = listSnapshots(suffix, cwd).at(-1);
+  if (!filePath) return null;
+  const body = fs.readFileSync(filePath, 'utf-8');
+  return { path: filePath, body, meta: parseFrontmatter(body) };
+}
+
 /**
  * Return the most recent snapshot for `slug`, or null. Polish reads this
  * to find its fix backlog when the slug matches.
  */
 export function readLatestSnapshot(slug, { cwd = process.cwd() } = {}) {
-  const all = listSnapshotsForSlug(slug, cwd);
-  if (!all.length) return null;
-  const latest = all[all.length - 1];
-  const body = fs.readFileSync(latest, 'utf-8');
-  return { path: latest, body, meta: parseFrontmatter(body) };
+  return readLatestSnapshotMatching(`__${slug}.md`, cwd);
+}
+
+/** Return the most recent snapshot across all targets, or null. */
+export function readLatestSnapshotAcrossTargets({ cwd = process.cwd() } = {}) {
+  return readLatestSnapshotMatching('.md', cwd);
 }
 
 /**
@@ -134,7 +143,7 @@ export function readLatestSnapshot(slug, { cwd = process.cwd() } = {}) {
  * Critique appends a one-line trend to its output using this.
  */
 export function readTrend(slug, { limit = 5, cwd = process.cwd() } = {}) {
-  const all = listSnapshotsForSlug(slug, cwd);
+  const all = listSnapshots(`__${slug}.md`, cwd);
   const slice = all.slice(-limit);
   return slice.map((file) => parseFrontmatter(fs.readFileSync(file, 'utf-8')));
 }

.claude/skills/impeccable/scripts/detector/browser/injected/index.mjs

@@ -683,6 +683,10 @@ if (IS_BROWSER) {
 
       const reasons = collectVisualContrastReasons(el, style);
       if (reasons.length === 0) continue;
+      // Image-only mode filters here, inside the cap: gradient/opacity/filter
+      // candidates earlier in DOM order must not consume the budget and
+      // starve the url()-backed texts this mode exists to sample.
+      if (options.imageOnly && !reasons.includes('image background')) continue;
 
       const textColor = parseRgb(style.color);
       const fontSize = parseFloat(style.fontSize) || 16;
@@ -1175,6 +1179,7 @@ if (IS_BROWSER) {
   }
 
   async function analyzeVisualContrast(options = {}) {
+    // imageOnly is enforced inside the collector, before the candidate cap.
     const candidates = collectVisualContrastCandidates(options);
     const results = [];
     const shouldScrollOffscreen = options.scrollOffscreen === true;
@@ -1260,9 +1265,16 @@ if (IS_BROWSER) {
 
   function addBrowserFindings(groupMap, el, findings) {
     if (!findings || findings.length === 0) return;
+    // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+    // matching findings for its whole subtree. Applied at this choke point so
+    // every per-element attribution (checks, layout, occlusion, rhythm)
+    // honors it; page-level findings attributed to <body> pass through
+    // untouched, since body has no ignoring ancestor.
+    const kept = findings.filter(f => !scopedIgnoreActive(el, f.type));
+    if (kept.length === 0) return;
     const existing = groupMap.get(el);
-    if (existing) existing.push(...findings);
-    else groupMap.set(el, [...findings]);
+    if (existing) existing.push(...kept);
+    else groupMap.set(el, [...kept]);
   }
 
   function browserFindingsFromMap(groupMap) {
@@ -1620,9 +1632,27 @@ if (IS_BROWSER) {
     for (const node of docClone.querySelectorAll('[id^="impeccable-live-"]')) {
       node.remove();
     }
-    const htmlPatternFindings = checkHtmlPatterns(docClone.outerHTML);
-    if (htmlPatternFindings.length > 0) {
-      const mapped = htmlPatternFindings.map(f => {
+    // Regex findings that name a live selector resolve against the real DOM:
+    // pseudo-element/class segments are stripped (the host element is the
+    // anchor), a selector that matches nothing on this page drops the finding
+    // (the CSS ships here, but the pattern never renders — the live DOM is
+    // ground truth in the browser), and a match under a data-impeccable-ignore
+    // ancestor is waived. Selector-less findings stay page-level.
+    const scopedHtmlFindings = checkHtmlPatterns(docClone.outerHTML).filter(f => {
+      if (!f.selector) return true;
+      const query = String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim().replace(/,\s*(?=,|$)/g, '');
+      if (!query || /^[,\s]*$/.test(query)) return true;
+      let matches;
+      try {
+        matches = document.querySelectorAll(query);
+      } catch {
+        return true;
+      }
+      if (matches.length === 0) return false;
+      return [...matches].some(el => !scopedIgnoreActive(el, f.id));
+    });
+    if (scopedHtmlFindings.length > 0) {
+      const mapped = scopedHtmlFindings.map(f => {
         const item = { type: f.id, detail: f.snippet };
         if (f.severity) {
           item.severity = f.severity;
@@ -1652,8 +1682,27 @@ if (IS_BROWSER) {
     };
   }
 
+  // Visual contrast has three modes. Explicit true runs the full sampled
+  // pass; explicit false disables it entirely (the deterministic-only mode
+  // the test suites use). Unset — the default overlay run — samples ONLY
+  // image-backed text: the one class the analytic walk deliberately skips,
+  // because a url() layer's pixels are unknowable without looking. In-page
+  // sampling draws the source image alone to a canvas (glyph ink never
+  // pollutes it), and a cross-origin image without CORS reports unresolved
+  // instead of guessing.
+  function visualContrastMode(options = {}) {
+    const explicit = typeof options.visualContrast === 'boolean'
+      ? options.visualContrast
+      : typeof window.__IMPECCABLE_CONFIG__?.visualContrast === 'boolean'
+        ? window.__IMPECCABLE_CONFIG__.visualContrast
+        : null;
+    if (explicit === true) return 'full';
+    if (explicit === false) return false;
+    return 'image-only';
+  }
+
   function shouldRunVisualContrast(options = {}) {
-    return options.visualContrast === true || window.__IMPECCABLE_CONFIG__?.visualContrast === true;
+    return visualContrastMode(options) !== false;
   }
 
   function visualContrastOptions(options = {}) {
@@ -1830,6 +1879,7 @@ if (IS_BROWSER) {
       return [];
     }
     const resolvedOptions = visualContrastOptions(options);
+    if (visualContrastMode(options) === 'image-only') resolvedOptions.imageOnly = true;
     const analyses = await analyzeVisualContrast(resolvedOptions);
     if (runtime.generation && runtime.generation !== scanGeneration) return analyses;
     lastVisualContrastAnalyses = analyses;

.claude/skills/impeccable/scripts/detector/detect-antipatterns-browser.js

@@ -861,6 +861,34 @@ function checkBorders(tag, widths, colors, radius, opts = {}) {
   return findings;
 }
 
+// ─── Scoped ignores: data-impeccable-ignore ─────────────────────────────────
+//
+// An element-scoped waiver that travels with the markup: any element carrying
+// `data-impeccable-ignore="rule-a rule-b"` (or `*`, or an empty value, for
+// every rule) suppresses matching findings from itself and its entire subtree,
+// in every engine that walks elements — the browser overlay, the extension,
+// and the static scan. This is the DOM twin of the line-based
+// `impeccable-disable` comment directives, which the browser cannot apply (a
+// live DOM has no line numbers), and the generalization of the one-off
+// `data-impeccable-allow-kickers` opt-out.
+//
+// The intended use is curated exhibits: a page that documents anti-patterns by
+// example, or renders a deliberate "before" specimen, marks the container once
+// and every engine skips it while still scanning the page around it.
+function scopedIgnoreActive(el, ruleId) {
+  const rule = String(ruleId || '').toLowerCase();
+  let cur = el;
+  while (cur && cur.nodeType === 1) {
+    const attr = typeof cur.getAttribute === 'function' ? cur.getAttribute('data-impeccable-ignore') : null;
+    if (attr != null) {
+      const rules = String(attr).trim().toLowerCase().split(/[\s,]+/).filter(Boolean);
+      if (rules.length === 0 || rules.includes('*') || rules.includes(rule)) return true;
+    }
+    cur = cur.parentElement;
+  }
+  return false;
+}
+
 // Returns true if the given text is composed entirely of emoji characters
 // (plus whitespace / variation selectors). Emojis render as multicolor glyphs
 // regardless of CSS `color`, so contrast checks against the element's text
@@ -1428,6 +1456,26 @@ function cssTextHasDarkRootBg(content, customProps) {
   return false;
 }
 
+// Best-effort extraction of the CSS selector whose declaration block contains
+// the given index in raw CSS text. Lets CSS-text findings carry a live-DOM
+// anchor, so the browser pass can resolve scoped ignores against the actual
+// element and drop patterns that render nowhere on the page. Returns null for
+// @-rule preludes, keyframe steps, nested blocks, and anything that does not
+// read as a selector; those findings stay page-level.
+function enclosingCssSelector(cssText, index) {
+  if (!cssText || !Number.isFinite(index)) return null;
+  const open = cssText.lastIndexOf('{', index);
+  if (open === -1) return null;
+  const prevClose = Math.max(cssText.lastIndexOf('}', open - 1), cssText.lastIndexOf(';', open - 1));
+  const raw = cssText.slice(prevClose + 1, open).trim().replace(/\s+/g, ' ');
+  if (!raw || raw.startsWith('@') || /^\d/.test(raw) || /[{}<]/.test(raw)) return null;
+  // Keyframe steps: percentage steps fail the digit test above, but `from`
+  // and `to` would read as (never-matching) type selectors and get a valid
+  // finding wrongly dropped by the zero-match rule downstream.
+  if (/^(?:from|to)(?:\s*,\s*(?:from|to))*$/i.test(raw)) return null;
+  return raw;
+}
+
 function scanCssTextForGlow(content) {
   const customProps = collectCssCustomProps(content);
   const hasDarkBg = cssTextHasDarkRootBg(content, customProps);
@@ -1739,6 +1787,7 @@ function scanCssTextForPseudoStripe(rawContent) {
       id: 'side-tab',
       snippet: `${selector} — absolute ${thicknessPx}px pseudo-element stripe (${edge}: 0)`,
       index: selectorStart,
+      selector,
     });
   }
   return findings;
@@ -1801,6 +1850,7 @@ function scanCssTextForInsetStripe(content) {
       findings.push({
         id: 'side-tab',
         snippet: `${selector} — inset box-shadow ${ay === 0 ? ax : ay}px stripe (${edge})`,
+        selector,
       });
       break;
     }
@@ -1858,7 +1908,7 @@ function collectMarqueeKeyframes(content) {
 function scanCssTextForMarquee(content, markup = content) {
   const findings = [];
   if (/<marquee\b/i.test(markup)) {
-    findings.push({ id: 'marquee', snippet: '<marquee> element' });
+    findings.push({ id: 'marquee', snippet: '<marquee> element', selector: 'marquee' });
   }
   const marqueeKeyframes = collectMarqueeKeyframes(content);
   if (marqueeKeyframes.size === 0) return findings;
@@ -1873,7 +1923,7 @@ function scanCssTextForMarquee(content, markup = content) {
       const key = `${selector} ${name}`;
       if (seen.has(key)) continue;
       seen.add(key);
-      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"` });
+      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"`, selector });
     }
   }
   return findings;
@@ -2244,8 +2294,10 @@ function checkHtmlPatterns(html, corpora) {
   const purpleHexRe = /#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9|6366f1|764ba2|667eea)\b/gi;
   if (purpleHexRe.test(styleText)) {
     const purpleTextRe = /(?:(?:^|;)\s*color\s*:\s*(?:.*?)(?:#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9))|gradient.*?#(?:7c3aed|8b5cf6|a855f7|764ba2|667eea))/gi;
-    if (purpleTextRe.test(styleText)) {
-      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected' });
+    purpleTextRe.lastIndex = 0;
+    const purpleMatch = purpleTextRe.exec(styleText);
+    if (purpleMatch) {
+      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected', selector: enclosingCssSelector(styleText, purpleMatch.index + 1) || undefined });
     }
   }
 
@@ -2256,7 +2308,7 @@ function checkHtmlPatterns(html, corpora) {
     const start = Math.max(0, gm.index - 200);
     const context = styleText.substring(start, gm.index + gm[0].length + 200);
     if (/gradient/i.test(context)) {
-      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient' });
+      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient', selector: enclosingCssSelector(styleText, gm.index) || undefined });
       break;
     }
   }
@@ -2322,7 +2374,7 @@ function checkHtmlPatterns(html, corpora) {
     const animationToken = bounceMatch[1]
       .split(/[,\s]+/)
       .find((part) => /bounce|elastic|wobble|jiggle|spring/i.test(part));
-    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}` });
+    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}`, selector: enclosingCssSelector(styleText, bounceMatch.index) || undefined });
   }
 
   // Overshoot cubic-bezier
@@ -2331,7 +2383,7 @@ function checkHtmlPatterns(html, corpora) {
   while ((bm = bezierRe.exec(styleText)) !== null) {
     const y1 = parseFloat(bm[2]), y2 = parseFloat(bm[4]);
     if (y1 < -0.1 || y1 > 1.1 || y2 < -0.1 || y2 > 1.1) {
-      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})` });
+      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})`, selector: enclosingCssSelector(styleText, bm.index) || undefined });
       break;
     }
   }
@@ -2364,18 +2416,21 @@ function checkHtmlPatterns(html, corpora) {
 
   const glowHits = scanCssTextForGlow(styleText);
   if (glowHits.length > 0) {
-    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet });
+    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet, selector: enclosingCssSelector(styleText, glowHits[0].index) || undefined });
   }
 
   // Radial-gradient background halo (gradient-drawn sibling of dark-glow)
   const haloHits = scanCssTextForRadialHalo(styleText);
   if (haloHits.length > 0) {
-    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet });
+    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet, selector: enclosingCssSelector(styleText, haloHits[0].index) || undefined });
   }
 
   // --- Generated-UI tells: repeating-gradient stripes ---
-  if (/repeating-(?:linear|radial|conic)-gradient\s*\(/i.test(styleText)) {
-    findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes' });
+  {
+    const stripesMatch = /repeating-(?:linear|radial|conic)-gradient\s*\(/i.exec(styleText);
+    if (stripesMatch) {
+      findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes', selector: enclosingCssSelector(styleText, stripesMatch.index) || undefined });
+    }
   }
 
   // --- Generated-UI tells: two-axis grid-line background ---
@@ -2393,7 +2448,7 @@ function checkHtmlPatterns(html, corpora) {
   // whole gradient layers.
   const gridHits = scanCssTextForGridBackground(styleText);
   if (gridHits.length > 0) {
-    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet });
+    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet, selector: enclosingCssSelector(styleText, gridHits[0].index) || undefined });
   }
 
   // --- Generated-copy tells: "X theater" framing copy ---
@@ -2413,8 +2468,11 @@ function checkHtmlPatterns(html, corpora) {
   // hover:rotate / hover:translate utility on an <img>. Each distinct
   // mechanism is its own finding.
   const imgHoverCss = /\bimg\b[^,{}]*:hover\b[^{}]*\{[^}]*\btransform\s*:\s*(?:scale|rotate|translate|matrix|skew)/i;
-  if (imgHoverCss.test(styleText)) {
-    findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule' });
+  {
+    const imgHoverMatch = imgHoverCss.exec(styleText);
+    if (imgHoverMatch) {
+      findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule', selector: enclosingCssSelector(styleText, imgHoverMatch.index + imgHoverMatch[0].indexOf('{') + 1) || undefined });
+    }
   }
   const imgTagRe = /<img\b[^>]*\bclass\s*=\s*"([^"]*)"/gi;
   let im;
@@ -2461,6 +2519,33 @@ function readOwnBackgroundColor(el, computedStyle) {
   return bg;
 }
 
+// One element's background-color as the cascade walk sees it: computed style
+// first (with the modern-color fallback), then, in static mode only,
+// custom-prop resolution and the inline-shorthand peek. Shared by
+// resolveBackground and resolveGradientStops so both walks read the same
+// surfaces.
+function readCascadeBackgroundColor(current, style, customPropMap) {
+  let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
+  if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
+    // The static engine can return literal "var(--X)" / "oklch(...)" strings.
+    // Resolve through customPropMap so Tailwind v4 color tokens become RGB.
+    if (customPropMap) {
+      bg = parseColorResolved(style.backgroundColor, customPropMap);
+    }
+    if (!bg || bg.a < 0.1) {
+      // Inline-style fallback for colors the static cascade did not surface
+      // on backgroundColor.
+      const rawStyle = current.getAttribute?.('style') || '';
+      const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
+      const inlineBg = bgMatch ? bgMatch[1].trim() : '';
+      if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
+        bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
+      }
+    }
+  }
+  return bg;
+}
+
 function resolveBackground(el, win, customPropMap) {
   let current = el;
   // Translucent layers (0.1 < a < 1) found on the way down to an opaque
@@ -2489,24 +2574,7 @@ function resolveBackground(el, win, customPropMap) {
     // body backgrounds.
     // Real browsers serialize wide-gamut computed values as oklab()/oklch()
     // (e.g. any color-mix() result), which plain parseRgb misses.
-    let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
-    if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
-      // jsdom returns literal "var(--X)" / "oklch(...)" strings. Resolve
-      // through customPropMap so Tailwind v4 color tokens become RGB.
-      if (customPropMap) {
-        bg = parseColorResolved(style.backgroundColor, customPropMap);
-      }
-      if (!bg || bg.a < 0.1) {
-        // Inline-style fallback. jsdom doesn't decompose background
-        // shorthand, so colors set via inline style are otherwise invisible.
-        const rawStyle = current.getAttribute?.('style') || '';
-        const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
-        const inlineBg = bgMatch ? bgMatch[1].trim() : '';
-        if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
-          bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
-        }
-      }
-    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
 
     if (bg && bg.a > 0.1) {
       if (bg.a >= 0.99) return flatten(bg);
@@ -2524,40 +2592,95 @@ function resolveBackground(el, win, customPropMap) {
     //   • on other elements: bail to null and let the caller fall back
     //     to gradient stops (gradient buttons / hero sections are real
     //     bgs worth checking against).
-    if (hasGradientOrUrl) {
-      if (current.tagName === 'BODY' || current.tagName === 'HTML') {
-        return flatten({ r: 255, g: 255, b: 255, a: 1 });
-      }
-      return null;
-    }
+    // A gradient or image with no solid color under it, at any level
+    // including body/html, means the visible ground is that layer itself.
+    // Return null so the caller measures against the actual gradient stops,
+    // or skips when nothing is parseable — skipping beats a wrong ratio.
+    //
+    // Body/html used to assume white here, a guard written for jsdom, which
+    // never decomposed the `background:` shorthand and so could not see the
+    // solid paper color a texture gradient usually sits on. It turned every
+    // light-on-dark page into a wall of low-contrast false positives (a dark
+    // oklch body gradient produced ~120 "on #ffffff" findings on one site).
+    // Both engines can see shorthand solids now — the browser natively, the
+    // static cascade via expandStaticDeclaration + var() resolution — so a
+    // missing solid is real, and the old failure case cannot recur: opaque
+    // stops fully cover any hidden solid (they ARE the ground), alpha stops
+    // composite over the resolved base or the white canvas default, and
+    // unresolvable stops drop rather than guess.
+    if (hasGradientOrUrl) return null;
     current = current.parentElement;
   }
   return flatten({ r: 255, g: 255, b: 255, a: 1 });
 }
 
+// parseGradientColors (shared) reads only the legacy serializations: rgb()
+// and hex stops. Browsers keep modern-space stops in computed backgroundImage
+// exactly as authored — `linear-gradient(oklch(7% 0.006 95), …)` stays oklch —
+// which is what every token-driven page produces. Route those through
+// parseAnyColor so a gradient ground is measurable rather than invisible.
+function parseGradientColorsModern(bgImage) {
+  if (!bgImage || !/gradient/i.test(bgImage)) return [];
+  const colors = parseGradientColors(bgImage);
+  for (const m of bgImage.matchAll(/(?:oklch|oklab|hsla?|hwb)\(\s*[^()]*\)/gi)) {
+    const c = parseAnyColor(m[0]);
+    if (c) colors.push(c);
+  }
+  return colors;
+}
+
 // Walk parents looking for a gradient background and return its color stops.
 // Used as a fallback when resolveBackground() returns null because the
 // effective background is a gradient (no single solid color to compare against).
+// Translucent solid layers found between the element and the gradient (frosted
+// panels, glass washes) are composited over every stop, the same way
+// resolveBackground flattens them over a solid base — raw stops alone would
+// false-flag dark text on a light frosted wash over a dark gradient, and miss
+// the inverse.
 function resolveGradientStops(el, win, customPropMap) {
   let current = el;
+  const overlays = [];
   while (current && current.nodeType === 1) {
     const style = DETECTOR_IS_BROWSER ? getComputedStyle(current) : win.getComputedStyle(current);
     const bgImage = style.backgroundImage || '';
+    // A url() layer anywhere in the stack — alone, or alongside a gradient in
+    // the same declaration (a translucent wash over a texture photo) — paints
+    // pixels the analytic walk cannot know. Measuring the gradient stops over
+    // the wrong base flagged dark ink sitting on a bright gold-leaf image at
+    // 2.6:1; skipping beats a wrong ratio, and the screenshot subsystem owns
+    // image-backed text.
+    if (bgImage && bgImage !== 'none' && /url\s*\(/i.test(bgImage)) return null;
     let stops = null;
     if (bgImage && bgImage !== 'none' && /gradient/i.test(bgImage)) {
-      const parsed = parseGradientColors(bgImage);
+      const parsed = parseGradientColorsModern(bgImage);
       if (parsed.length > 0) stops = parsed;
     }
     if (!stops && !DETECTOR_IS_BROWSER) {
-      // jsdom doesn't decompose `background:` shorthand — peek at the raw inline style
+      // Static mode: peek at the raw inline style for gradients the cascade did not surface
       const rawStyle = current.getAttribute?.('style') || '';
       const bgMatch = rawStyle.match(/background(?:-image)?\s*:\s*([^;]+)/i);
       if (bgMatch && /gradient/i.test(bgMatch[1])) {
-        const parsed = parseGradientColors(bgMatch[1]);
+        const parsed = parseGradientColorsModern(bgMatch[1]);
         if (parsed.length > 0) stops = parsed;
       }
     }
-    if (stops) return compositeGradientStops(stops, current, win, customPropMap);
+    if (stops) {
+      const composited = compositeGradientStops(stops, current, win, customPropMap);
+      if (!composited || overlays.length === 0) return composited;
+      return composited.map(stop => {
+        let acc = stop;
+        for (let i = overlays.length - 1; i >= 0; i--) acc = compositeColorOver(overlays[i], acc);
+        return acc;
+      });
+    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
+    if (bg && bg.a > 0.1) {
+      // An opaque surface above the gradient means the gradient never shows
+      // through here; resolveBackground would have returned it, so reaching
+      // this is defensive — bail rather than measure the wrong layer.
+      if (bg.a >= 0.99) return null;
+      overlays.push(bg);
+    }
     current = current.parentElement;
   }
   return null;
@@ -2777,6 +2900,10 @@ function checkElementColorsDOM(el) {
   const rect = el.getBoundingClientRect();
   if (rect.width < 10 || rect.height < 10) return [];
   const style = getComputedStyle(el);
+  // Invisible at rest: hidden scene variants (opacity-0 carousels, swap
+  // decks) are not user-visible, and measuring their inherited colors against
+  // whatever surface happens to sit behind the stack is noise, not audit.
+  if (style.visibility === 'hidden' || effectiveOpacityDOM(el) <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
   let effectiveBg = resolveBackground(el);
@@ -3589,11 +3716,13 @@ function checkElementGlowDOM(el) {
   // If resolveBackground returns null (gradient), try to infer from the gradient colors
   let parentBg = el.parentElement ? resolveBackground(el.parentElement) : resolveBackground(el);
   if (!parentBg) {
-    // Gradient background — sample its colors to determine if it's dark
+    // Gradient background — sample its colors to determine if it's dark.
+    // Modern-syntax parsing matters here: body-level gradients now reach this
+    // fallback in browser mode, and their stops usually serialize as oklch.
     let cur = el.parentElement;
     while (cur && cur.nodeType === 1) {
       const bgImage = getComputedStyle(cur).backgroundImage || '';
-      const gradColors = parseGradientColors(bgImage);
+      const gradColors = parseGradientColorsModern(bgImage);
       if (gradColors.length > 0) {
         // Average the gradient colors
         const avg = { r: 0, g: 0, b: 0 };
@@ -4435,6 +4564,14 @@ function checkElementBorders(tag, style, overrides, resolvedRadius, el = null) {
 }
 
 function checkElementColors(el, style, tag, window, customPropMap, hasAnchorInheritRule) {
+  // Invisible at rest, static twin of the browser walk's skip: opacity does
+  // not inherit, so walk ancestors multiplying declared opacity down.
+  if (style.visibility === 'hidden') return [];
+  let effOpacity = 1;
+  for (let cur = el; cur && cur.nodeType === 1 && effOpacity > 0.02; cur = cur.parentElement) {
+    effOpacity *= parseFloat(window.getComputedStyle(cur).opacity || '1');
+  }
+  if (effOpacity <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
 
@@ -5593,6 +5730,11 @@ function isRenderedForBrowserRule(el) {
 function checkElementTextOverflowDOM(el) {
   const tag = el.tagName.toLowerCase();
   if (TEXT_OVERFLOW_SKIP_TAGS.has(tag)) return [];
+  // scrollWidth/clientWidth are CSS box-model metrics; on SVG content Chrome
+  // returns arbitrary non-zero values for both (a <text> reported 78/48 while
+  // its rendered length sat comfortably inside its box), so the delta is
+  // noise, not overflow. SVG clips to its own viewport anyway.
+  if (el.namespaceURI === 'http://www.w3.org/2000/svg') return [];
   if (!isRenderedForBrowserRule(el)) return [];
   // Only the element that actually owns overflowing text — not its ancestors,
   // which inherit a wider scrollWidth from the spilling descendant.
@@ -5977,6 +6119,22 @@ function isPaintedForOcclusion(el) {
 // path is pure geometry and runs anywhere on the page.
 const OCCLUSION_TEXT_SKIP_TAGS = new Set(['script', 'style', 'noscript', 'template', 'title']);
 
+// An element whose effective opacity multiplies out to ~0 paints nothing at
+// rest: it is not user-visible, so visual findings on it (contrast, occlusion)
+// measure a state nobody sees. Browser-only — the walk needs live computed
+// styles. Cycling scenes that fade such elements in later are the screenshot
+// subsystem's territory, not the analytic walk's.
+function effectiveOpacityDOM(el) {
+  let o = 1;
+  // Walk all the way through body and html: `body { opacity: 0 }` page-fade
+  // wrappers hide every descendant just as thoroughly as a local wrapper.
+  for (let cur = el; cur && cur.nodeType === 1; cur = cur.parentElement) {
+    o *= parseFloat(getComputedStyle(cur).opacity || '1');
+    if (o <= 0.02) return 0;
+  }
+  return o;
+}
+
 function checkTextOcclusionDOM() {
   const findings = [];
   const seenVictims = new Set();
@@ -6004,6 +6162,11 @@ function checkTextOcclusionDOM() {
     }
     return false;
   };
+  // The classic occluder shape this rules out is an opacity-0 interaction
+  // layer — a range scrubber stretched over a before/after comparison — which
+  // elementFromPoint still returns and whose UA background-color otherwise
+  // reads as an opaque box.
+  const effectiveOpacity = effectiveOpacityDOM;
 
   // Collect renderable text owners in / near the first viewport for the
   // elementFromPoint probe. SVG <text> counts too.
@@ -6016,6 +6179,7 @@ function checkTextOcclusionDOM() {
     const text = inSvg ? (el.textContent || '').trim() : elementDirectText(el);
     if (text.length < 2) continue;
     if (!isPaintedForOcclusion(el)) continue;
+    if (effectiveOpacity(el) <= 0.02) continue;
     let rect; try { rect = el.getBoundingClientRect(); } catch { continue; }
     if (rect.width < 6 || rect.height < 6) continue;
     // Viewport-bound probe: keep text whose box overlaps the live viewport.
@@ -6049,6 +6213,7 @@ function checkTextOcclusionDOM() {
         if (top === el || el.contains(top) || top.contains(el)) continue;
         const topCs = getComputedStyle(top);
         if (isFloated(topCs) || isMarqueeish(top, topCs) || isPinnedOverlay(top)) continue;
+        if (effectiveOpacity(top) <= 0.02) continue;
         const topTag = top.tagName.toLowerCase();
         // Text sitting under a raw image/video is contrast territory (deduped
         // against the pixel low-contrast rule); leave those alone here.
@@ -6941,6 +7106,10 @@ if (IS_BROWSER) {
 
       const reasons = collectVisualContrastReasons(el, style);
       if (reasons.length === 0) continue;
+      // Image-only mode filters here, inside the cap: gradient/opacity/filter
+      // candidates earlier in DOM order must not consume the budget and
+      // starve the url()-backed texts this mode exists to sample.
+      if (options.imageOnly && !reasons.includes('image background')) continue;
 
       const textColor = parseRgb(style.color);
       const fontSize = parseFloat(style.fontSize) || 16;
@@ -7433,6 +7602,7 @@ if (IS_BROWSER) {
   }
 
   async function analyzeVisualContrast(options = {}) {
+    // imageOnly is enforced inside the collector, before the candidate cap.
     const candidates = collectVisualContrastCandidates(options);
     const results = [];
     const shouldScrollOffscreen = options.scrollOffscreen === true;
@@ -7518,9 +7688,16 @@ if (IS_BROWSER) {
 
   function addBrowserFindings(groupMap, el, findings) {
     if (!findings || findings.length === 0) return;
+    // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+    // matching findings for its whole subtree. Applied at this choke point so
+    // every per-element attribution (checks, layout, occlusion, rhythm)
+    // honors it; page-level findings attributed to <body> pass through
+    // untouched, since body has no ignoring ancestor.
+    const kept = findings.filter(f => !scopedIgnoreActive(el, f.type));
+    if (kept.length === 0) return;
     const existing = groupMap.get(el);
-    if (existing) existing.push(...findings);
-    else groupMap.set(el, [...findings]);
+    if (existing) existing.push(...kept);
+    else groupMap.set(el, [...kept]);
   }
 
   function browserFindingsFromMap(groupMap) {
@@ -7878,9 +8055,27 @@ if (IS_BROWSER) {
     for (const node of docClone.querySelectorAll('[id^="impeccable-live-"]')) {
       node.remove();
     }
-    const htmlPatternFindings = checkHtmlPatterns(docClone.outerHTML);
-    if (htmlPatternFindings.length > 0) {
-      const mapped = htmlPatternFindings.map(f => {
+    // Regex findings that name a live selector resolve against the real DOM:
+    // pseudo-element/class segments are stripped (the host element is the
+    // anchor), a selector that matches nothing on this page drops the finding
+    // (the CSS ships here, but the pattern never renders — the live DOM is
+    // ground truth in the browser), and a match under a data-impeccable-ignore
+    // ancestor is waived. Selector-less findings stay page-level.
+    const scopedHtmlFindings = checkHtmlPatterns(docClone.outerHTML).filter(f => {
+      if (!f.selector) return true;
+      const query = String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim().replace(/,\s*(?=,|$)/g, '');
+      if (!query || /^[,\s]*$/.test(query)) return true;
+      let matches;
+      try {
+        matches = document.querySelectorAll(query);
+      } catch {
+        return true;
+      }
+      if (matches.length === 0) return false;
+      return [...matches].some(el => !scopedIgnoreActive(el, f.id));
+    });
+    if (scopedHtmlFindings.length > 0) {
+      const mapped = scopedHtmlFindings.map(f => {
         const item = { type: f.id, detail: f.snippet };
         if (f.severity) {
           item.severity = f.severity;
@@ -7910,8 +8105,27 @@ if (IS_BROWSER) {
     };
   }
 
+  // Visual contrast has three modes. Explicit true runs the full sampled
+  // pass; explicit false disables it entirely (the deterministic-only mode
+  // the test suites use). Unset — the default overlay run — samples ONLY
+  // image-backed text: the one class the analytic walk deliberately skips,
+  // because a url() layer's pixels are unknowable without looking. In-page
+  // sampling draws the source image alone to a canvas (glyph ink never
+  // pollutes it), and a cross-origin image without CORS reports unresolved
+  // instead of guessing.
+  function visualContrastMode(options = {}) {
+    const explicit = typeof options.visualContrast === 'boolean'
+      ? options.visualContrast
+      : typeof window.__IMPECCABLE_CONFIG__?.visualContrast === 'boolean'
+        ? window.__IMPECCABLE_CONFIG__.visualContrast
+        : null;
+    if (explicit === true) return 'full';
+    if (explicit === false) return false;
+    return 'image-only';
+  }
+
   function shouldRunVisualContrast(options = {}) {
-    return options.visualContrast === true || window.__IMPECCABLE_CONFIG__?.visualContrast === true;
+    return visualContrastMode(options) !== false;
   }
 
   function visualContrastOptions(options = {}) {
@@ -8088,6 +8302,7 @@ if (IS_BROWSER) {
       return [];
     }
     const resolvedOptions = visualContrastOptions(options);
+    if (visualContrastMode(options) === 'image-only') resolvedOptions.imageOnly = true;
     const analyses = await analyzeVisualContrast(resolvedOptions);
     if (runtime.generation && runtime.generation !== scanGeneration) return analyses;
     lastVisualContrastAnalyses = analyses;

.claude/skills/impeccable/scripts/detector/detect-antipatterns.mjs

@@ -35,6 +35,7 @@ export { detectUrl, createBrowserDetector } from './engines/browser/detect-url.m
 export { detectText, extractStyleBlocks, extractCSSinJS } from './engines/regex/detect-text.mjs';
 export {
   walkDir,
+  hasScannableExtension,
   SCANNABLE_EXTENSIONS,
   SKIP_DIRS,
   buildImportGraph,

.claude/skills/impeccable/scripts/detector/engines/static-html/css-cascade.mjs

@@ -226,6 +226,10 @@ const STATIC_INHERITED_PROPS = new Set([
   'color', 'fontFamily', 'fontSize', 'fontStyle', 'fontWeight', 'fontVariant',
   'lineHeight', 'letterSpacing', 'textTransform', 'textAlign', 'hyphens',
   'webkitHyphens',
+  // visibility inherits in real CSS, and the invisible-at-rest contrast skip
+  // relies on descendants of a hidden container computing as hidden. A child
+  // that declares `visibility: visible` still overrides the inherited value.
+  'visibility',
 ]);
 
 const STATIC_DEFAULT_STYLE = {
@@ -278,6 +282,7 @@ const STATIC_DEFAULT_STYLE = {
   marginLeft: '0px',
   position: 'static',
   visibility: 'visible',
+  opacity: '1',
   top: 'auto',
   right: 'auto',
   bottom: 'auto',
@@ -334,6 +339,7 @@ const STATIC_PROP_MAP = {
   'margin-left': 'marginLeft',
   'position': 'position',
   'visibility': 'visibility',
+  'opacity': 'opacity',
   'top': 'top',
   'right': 'right',
   'bottom': 'bottom',

.claude/skills/impeccable/scripts/detector/engines/static-html/detect-html.mjs

@@ -28,6 +28,7 @@ import {
   checkCreamPalette,
   checkHtmlPatterns,
   checkKickerAboveHeadingFromDoc,
+  scopedIgnoreActive,
   checkNumberedSectionLabelsFromDoc,
   checkPageLayout,
   checkPageQualityFromDoc,
@@ -138,10 +139,21 @@ async function detectHtml(filePath, options = {}) {
         domutils,
       };
     });
-  } catch {
-    return detectText(html, filePath, options);
+  } catch (err) {
+  if (!globalThis.__impeccableStaticHtmlWarned) {
+    globalThis.__impeccableStaticHtmlWarned = true;
+
+    process.stderr.write(
+  'impeccable detect: DEGRADED - HTML parser modules unavailable ' +
+  '(htmlparser2, css-select, css-tree, domutils).\n' +
+  'Falling back to regex matching. Custom properties, selector matching and computed ' +
+  'contrast are NOT evaluated; findings are an undercount, not a clean bill of health.\n'
+);
   }
 
+  return detectText(html, filePath, options);
+}
+
   const resolvedPath = path.resolve(filePath);
   const fileDir = path.dirname(resolvedPath);
   const root = profileStep(profile, {
@@ -171,6 +183,9 @@ async function detectHtml(filePath, options = {}) {
       const tag = el.tagName.toLowerCase();
       const style = window.getComputedStyle(el);
       for (const f of runElementCheck(rule.id, () => rule.run(el, tag, style, window, customPropMap))) {
+        // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+        // matching findings for its subtree, same as the browser walk.
+        if (scopedIgnoreActive(el, f.id)) continue;
         findings.push(finding(f.id, filePath, f.snippet));
       }
     }
@@ -238,6 +253,17 @@ async function detectHtml(filePath, options = {}) {
     for (const f of runPageCheck('html-patterns', () => checkHtmlPatterns(html, patternCorpora).filter(item =>
       item.id !== 'bounce-easing' && item.id !== 'layout-transition'
     ))) {
+      // Selector-backed page findings honor scoped waivers here too, matching
+      // the browser pass: resolve the selector and drop the finding when an
+      // ignoring ancestor covers a match. Unlike the browser, an unmatched
+      // selector keeps the finding — static scans see partial documents.
+      if (f.selector) {
+        let matches = null;
+        try {
+          matches = document.querySelectorAll(String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim());
+        } catch { matches = null; }
+        if (matches && matches.length > 0 && [...matches].every(el => scopedIgnoreActive(el, f.id))) continue;
+      }
       const item = finding(f.id, filePath, f.snippet);
       // Position-aware severity promotion: checks may attach a per-finding
       // severity (e.g. a pulsing dot inside a header/nav landmark) that

.claude/skills/impeccable/scripts/detector/node/file-system.mjs

@@ -26,11 +26,20 @@ const HIDDEN_SOURCE_DIRS = new Set(['.vitepress', '.vuepress', '.storybook']);
 const SCANNABLE_EXTENSIONS = new Set([
   '.html', '.htm', '.css', '.scss', '.sass', '.less',
   '.jsx', '.tsx', '.js', '.ts',
-  '.vue', '.svelte', '.astro',
+  '.vue', '.svelte', '.astro', '.blade.php',
 ]);
 
 const HTML_EXTENSIONS = new Set(['.html', '.htm']);
 
+function hasScannableExtension(filename) {
+  const lower = filename.toLowerCase();
+  if (SCANNABLE_EXTENSIONS.has(path.extname(lower))) return true;
+  for (const ext of SCANNABLE_EXTENSIONS) {
+    if (ext.indexOf('.', 1) !== -1 && lower.endsWith(ext)) return true;
+  }
+  return false;
+}
+
 const IMPORT_SPECIFIER_PATTERNS = [
   /import\s+(?:[\s\S]*?from\s+)?['"]([^'"]+)['"]/g,
   /@import\s+(?:url\(\s*)?['"]?([^'");\s]+)['"]?\s*\)?/g,
@@ -46,7 +55,7 @@ function walkDir(dir) {
     if (entry.isDirectory() && entry.name.startsWith('.') && !HIDDEN_SOURCE_DIRS.has(entry.name)) continue;
     const full = path.join(dir, entry.name);
     if (entry.isDirectory()) files.push(...walkDir(full));
-    else if (SCANNABLE_EXTENSIONS.has(path.extname(entry.name).toLowerCase())) files.push(full);
+    else if (hasScannableExtension(entry.name)) files.push(full);
   }
   return files;
 }
@@ -194,6 +203,7 @@ export {
   SKIP_DIRS,
   SCANNABLE_EXTENSIONS,
   HTML_EXTENSIONS,
+  hasScannableExtension,
   walkDir,
   resolveImport,
   buildImportGraph,

.claude/skills/impeccable/scripts/detector/rules/checks.mjs

@@ -70,6 +70,34 @@ function checkBorders(tag, widths, colors, radius, opts = {}) {
   return findings;
 }
 
+// ─── Scoped ignores: data-impeccable-ignore ─────────────────────────────────
+//
+// An element-scoped waiver that travels with the markup: any element carrying
+// `data-impeccable-ignore="rule-a rule-b"` (or `*`, or an empty value, for
+// every rule) suppresses matching findings from itself and its entire subtree,
+// in every engine that walks elements — the browser overlay, the extension,
+// and the static scan. This is the DOM twin of the line-based
+// `impeccable-disable` comment directives, which the browser cannot apply (a
+// live DOM has no line numbers), and the generalization of the one-off
+// `data-impeccable-allow-kickers` opt-out.
+//
+// The intended use is curated exhibits: a page that documents anti-patterns by
+// example, or renders a deliberate "before" specimen, marks the container once
+// and every engine skips it while still scanning the page around it.
+function scopedIgnoreActive(el, ruleId) {
+  const rule = String(ruleId || '').toLowerCase();
+  let cur = el;
+  while (cur && cur.nodeType === 1) {
+    const attr = typeof cur.getAttribute === 'function' ? cur.getAttribute('data-impeccable-ignore') : null;
+    if (attr != null) {
+      const rules = String(attr).trim().toLowerCase().split(/[\s,]+/).filter(Boolean);
+      if (rules.length === 0 || rules.includes('*') || rules.includes(rule)) return true;
+    }
+    cur = cur.parentElement;
+  }
+  return false;
+}
+
 // Returns true if the given text is composed entirely of emoji characters
 // (plus whitespace / variation selectors). Emojis render as multicolor glyphs
 // regardless of CSS `color`, so contrast checks against the element's text
@@ -637,6 +665,26 @@ function cssTextHasDarkRootBg(content, customProps) {
   return false;
 }
 
+// Best-effort extraction of the CSS selector whose declaration block contains
+// the given index in raw CSS text. Lets CSS-text findings carry a live-DOM
+// anchor, so the browser pass can resolve scoped ignores against the actual
+// element and drop patterns that render nowhere on the page. Returns null for
+// @-rule preludes, keyframe steps, nested blocks, and anything that does not
+// read as a selector; those findings stay page-level.
+function enclosingCssSelector(cssText, index) {
+  if (!cssText || !Number.isFinite(index)) return null;
+  const open = cssText.lastIndexOf('{', index);
+  if (open === -1) return null;
+  const prevClose = Math.max(cssText.lastIndexOf('}', open - 1), cssText.lastIndexOf(';', open - 1));
+  const raw = cssText.slice(prevClose + 1, open).trim().replace(/\s+/g, ' ');
+  if (!raw || raw.startsWith('@') || /^\d/.test(raw) || /[{}<]/.test(raw)) return null;
+  // Keyframe steps: percentage steps fail the digit test above, but `from`
+  // and `to` would read as (never-matching) type selectors and get a valid
+  // finding wrongly dropped by the zero-match rule downstream.
+  if (/^(?:from|to)(?:\s*,\s*(?:from|to))*$/i.test(raw)) return null;
+  return raw;
+}
+
 function scanCssTextForGlow(content) {
   const customProps = collectCssCustomProps(content);
   const hasDarkBg = cssTextHasDarkRootBg(content, customProps);
@@ -948,6 +996,7 @@ function scanCssTextForPseudoStripe(rawContent) {
       id: 'side-tab',
       snippet: `${selector} — absolute ${thicknessPx}px pseudo-element stripe (${edge}: 0)`,
       index: selectorStart,
+      selector,
     });
   }
   return findings;
@@ -1010,6 +1059,7 @@ function scanCssTextForInsetStripe(content) {
       findings.push({
         id: 'side-tab',
         snippet: `${selector} — inset box-shadow ${ay === 0 ? ax : ay}px stripe (${edge})`,
+        selector,
       });
       break;
     }
@@ -1067,7 +1117,7 @@ function collectMarqueeKeyframes(content) {
 function scanCssTextForMarquee(content, markup = content) {
   const findings = [];
   if (/<marquee\b/i.test(markup)) {
-    findings.push({ id: 'marquee', snippet: '<marquee> element' });
+    findings.push({ id: 'marquee', snippet: '<marquee> element', selector: 'marquee' });
   }
   const marqueeKeyframes = collectMarqueeKeyframes(content);
   if (marqueeKeyframes.size === 0) return findings;
@@ -1082,7 +1132,7 @@ function scanCssTextForMarquee(content, markup = content) {
       const key = `${selector} ${name}`;
       if (seen.has(key)) continue;
       seen.add(key);
-      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"` });
+      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"`, selector });
     }
   }
   return findings;
@@ -1453,8 +1503,10 @@ function checkHtmlPatterns(html, corpora) {
   const purpleHexRe = /#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9|6366f1|764ba2|667eea)\b/gi;
   if (purpleHexRe.test(styleText)) {
     const purpleTextRe = /(?:(?:^|;)\s*color\s*:\s*(?:.*?)(?:#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9))|gradient.*?#(?:7c3aed|8b5cf6|a855f7|764ba2|667eea))/gi;
-    if (purpleTextRe.test(styleText)) {
-      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected' });
+    purpleTextRe.lastIndex = 0;
+    const purpleMatch = purpleTextRe.exec(styleText);
+    if (purpleMatch) {
+      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected', selector: enclosingCssSelector(styleText, purpleMatch.index + 1) || undefined });
     }
   }
 
@@ -1465,7 +1517,7 @@ function checkHtmlPatterns(html, corpora) {
     const start = Math.max(0, gm.index - 200);
     const context = styleText.substring(start, gm.index + gm[0].length + 200);
     if (/gradient/i.test(context)) {
-      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient' });
+      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient', selector: enclosingCssSelector(styleText, gm.index) || undefined });
       break;
     }
   }
@@ -1531,7 +1583,7 @@ function checkHtmlPatterns(html, corpora) {
     const animationToken = bounceMatch[1]
       .split(/[,\s]+/)
       .find((part) => /bounce|elastic|wobble|jiggle|spring/i.test(part));
-    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}` });
+    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}`, selector: enclosingCssSelector(styleText, bounceMatch.index) || undefined });
   }
 
   // Overshoot cubic-bezier
@@ -1540,7 +1592,7 @@ function checkHtmlPatterns(html, corpora) {
   while ((bm = bezierRe.exec(styleText)) !== null) {
     const y1 = parseFloat(bm[2]), y2 = parseFloat(bm[4]);
     if (y1 < -0.1 || y1 > 1.1 || y2 < -0.1 || y2 > 1.1) {
-      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})` });
+      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})`, selector: enclosingCssSelector(styleText, bm.index) || undefined });
       break;
     }
   }
@@ -1573,18 +1625,21 @@ function checkHtmlPatterns(html, corpora) {
 
   const glowHits = scanCssTextForGlow(styleText);
   if (glowHits.length > 0) {
-    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet });
+    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet, selector: enclosingCssSelector(styleText, glowHits[0].index) || undefined });
   }
 
   // Radial-gradient background halo (gradient-drawn sibling of dark-glow)
   const haloHits = scanCssTextForRadialHalo(styleText);
   if (haloHits.length > 0) {
-    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet });
+    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet, selector: enclosingCssSelector(styleText, haloHits[0].index) || undefined });
   }
 
   // --- Generated-UI tells: repeating-gradient stripes ---
-  if (/repeating-(?:linear|radial|conic)-gradient\s*\(/i.test(styleText)) {
-    findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes' });
+  {
+    const stripesMatch = /repeating-(?:linear|radial|conic)-gradient\s*\(/i.exec(styleText);
+    if (stripesMatch) {
+      findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes', selector: enclosingCssSelector(styleText, stripesMatch.index) || undefined });
+    }
   }
 
   // --- Generated-UI tells: two-axis grid-line background ---
@@ -1602,7 +1657,7 @@ function checkHtmlPatterns(html, corpora) {
   // whole gradient layers.
   const gridHits = scanCssTextForGridBackground(styleText);
   if (gridHits.length > 0) {
-    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet });
+    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet, selector: enclosingCssSelector(styleText, gridHits[0].index) || undefined });
   }
 
   // --- Generated-copy tells: "X theater" framing copy ---
@@ -1622,8 +1677,11 @@ function checkHtmlPatterns(html, corpora) {
   // hover:rotate / hover:translate utility on an <img>. Each distinct
   // mechanism is its own finding.
   const imgHoverCss = /\bimg\b[^,{}]*:hover\b[^{}]*\{[^}]*\btransform\s*:\s*(?:scale|rotate|translate|matrix|skew)/i;
-  if (imgHoverCss.test(styleText)) {
-    findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule' });
+  {
+    const imgHoverMatch = imgHoverCss.exec(styleText);
+    if (imgHoverMatch) {
+      findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule', selector: enclosingCssSelector(styleText, imgHoverMatch.index + imgHoverMatch[0].indexOf('{') + 1) || undefined });
+    }
   }
   const imgTagRe = /<img\b[^>]*\bclass\s*=\s*"([^"]*)"/gi;
   let im;
@@ -1670,6 +1728,33 @@ function readOwnBackgroundColor(el, computedStyle) {
   return bg;
 }
 
+// One element's background-color as the cascade walk sees it: computed style
+// first (with the modern-color fallback), then, in static mode only,
+// custom-prop resolution and the inline-shorthand peek. Shared by
+// resolveBackground and resolveGradientStops so both walks read the same
+// surfaces.
+function readCascadeBackgroundColor(current, style, customPropMap) {
+  let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
+  if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
+    // The static engine can return literal "var(--X)" / "oklch(...)" strings.
+    // Resolve through customPropMap so Tailwind v4 color tokens become RGB.
+    if (customPropMap) {
+      bg = parseColorResolved(style.backgroundColor, customPropMap);
+    }
+    if (!bg || bg.a < 0.1) {
+      // Inline-style fallback for colors the static cascade did not surface
+      // on backgroundColor.
+      const rawStyle = current.getAttribute?.('style') || '';
+      const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
+      const inlineBg = bgMatch ? bgMatch[1].trim() : '';
+      if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
+        bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
+      }
+    }
+  }
+  return bg;
+}
+
 function resolveBackground(el, win, customPropMap) {
   let current = el;
   // Translucent layers (0.1 < a < 1) found on the way down to an opaque
@@ -1698,24 +1783,7 @@ function resolveBackground(el, win, customPropMap) {
     // body backgrounds.
     // Real browsers serialize wide-gamut computed values as oklab()/oklch()
     // (e.g. any color-mix() result), which plain parseRgb misses.
-    let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
-    if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
-      // jsdom returns literal "var(--X)" / "oklch(...)" strings. Resolve
-      // through customPropMap so Tailwind v4 color tokens become RGB.
-      if (customPropMap) {
-        bg = parseColorResolved(style.backgroundColor, customPropMap);
-      }
-      if (!bg || bg.a < 0.1) {
-        // Inline-style fallback. jsdom doesn't decompose background
-        // shorthand, so colors set via inline style are otherwise invisible.
-        const rawStyle = current.getAttribute?.('style') || '';
-        const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
-        const inlineBg = bgMatch ? bgMatch[1].trim() : '';
-        if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
-          bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
-        }
-      }
-    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
 
     if (bg && bg.a > 0.1) {
       if (bg.a >= 0.99) return flatten(bg);
@@ -1733,40 +1801,95 @@ function resolveBackground(el, win, customPropMap) {
     //   • on other elements: bail to null and let the caller fall back
     //     to gradient stops (gradient buttons / hero sections are real
     //     bgs worth checking against).
-    if (hasGradientOrUrl) {
-      if (current.tagName === 'BODY' || current.tagName === 'HTML') {
-        return flatten({ r: 255, g: 255, b: 255, a: 1 });
-      }
-      return null;
-    }
+    // A gradient or image with no solid color under it, at any level
+    // including body/html, means the visible ground is that layer itself.
+    // Return null so the caller measures against the actual gradient stops,
+    // or skips when nothing is parseable — skipping beats a wrong ratio.
+    //
+    // Body/html used to assume white here, a guard written for jsdom, which
+    // never decomposed the `background:` shorthand and so could not see the
+    // solid paper color a texture gradient usually sits on. It turned every
+    // light-on-dark page into a wall of low-contrast false positives (a dark
+    // oklch body gradient produced ~120 "on #ffffff" findings on one site).
+    // Both engines can see shorthand solids now — the browser natively, the
+    // static cascade via expandStaticDeclaration + var() resolution — so a
+    // missing solid is real, and the old failure case cannot recur: opaque
+    // stops fully cover any hidden solid (they ARE the ground), alpha stops
+    // composite over the resolved base or the white canvas default, and
+    // unresolvable stops drop rather than guess.
+    if (hasGradientOrUrl) return null;
     current = current.parentElement;
   }
   return flatten({ r: 255, g: 255, b: 255, a: 1 });
 }
 
+// parseGradientColors (shared) reads only the legacy serializations: rgb()
+// and hex stops. Browsers keep modern-space stops in computed backgroundImage
+// exactly as authored — `linear-gradient(oklch(7% 0.006 95), …)` stays oklch —
+// which is what every token-driven page produces. Route those through
+// parseAnyColor so a gradient ground is measurable rather than invisible.
+function parseGradientColorsModern(bgImage) {
+  if (!bgImage || !/gradient/i.test(bgImage)) return [];
+  const colors = parseGradientColors(bgImage);
+  for (const m of bgImage.matchAll(/(?:oklch|oklab|hsla?|hwb)\(\s*[^()]*\)/gi)) {
+    const c = parseAnyColor(m[0]);
+    if (c) colors.push(c);
+  }
+  return colors;
+}
+
 // Walk parents looking for a gradient background and return its color stops.
 // Used as a fallback when resolveBackground() returns null because the
 // effective background is a gradient (no single solid color to compare against).
+// Translucent solid layers found between the element and the gradient (frosted
+// panels, glass washes) are composited over every stop, the same way
+// resolveBackground flattens them over a solid base — raw stops alone would
+// false-flag dark text on a light frosted wash over a dark gradient, and miss
+// the inverse.
 function resolveGradientStops(el, win, customPropMap) {
   let current = el;
+  const overlays = [];
   while (current && current.nodeType === 1) {
     const style = DETECTOR_IS_BROWSER ? getComputedStyle(current) : win.getComputedStyle(current);
     const bgImage = style.backgroundImage || '';
+    // A url() layer anywhere in the stack — alone, or alongside a gradient in
+    // the same declaration (a translucent wash over a texture photo) — paints
+    // pixels the analytic walk cannot know. Measuring the gradient stops over
+    // the wrong base flagged dark ink sitting on a bright gold-leaf image at
+    // 2.6:1; skipping beats a wrong ratio, and the screenshot subsystem owns
+    // image-backed text.
+    if (bgImage && bgImage !== 'none' && /url\s*\(/i.test(bgImage)) return null;
     let stops = null;
     if (bgImage && bgImage !== 'none' && /gradient/i.test(bgImage)) {
-      const parsed = parseGradientColors(bgImage);
+      const parsed = parseGradientColorsModern(bgImage);
       if (parsed.length > 0) stops = parsed;
     }
     if (!stops && !DETECTOR_IS_BROWSER) {
-      // jsdom doesn't decompose `background:` shorthand — peek at the raw inline style
+      // Static mode: peek at the raw inline style for gradients the cascade did not surface
       const rawStyle = current.getAttribute?.('style') || '';
       const bgMatch = rawStyle.match(/background(?:-image)?\s*:\s*([^;]+)/i);
       if (bgMatch && /gradient/i.test(bgMatch[1])) {
-        const parsed = parseGradientColors(bgMatch[1]);
+        const parsed = parseGradientColorsModern(bgMatch[1]);
         if (parsed.length > 0) stops = parsed;
       }
     }
-    if (stops) return compositeGradientStops(stops, current, win, customPropMap);
+    if (stops) {
+      const composited = compositeGradientStops(stops, current, win, customPropMap);
+      if (!composited || overlays.length === 0) return composited;
+      return composited.map(stop => {
+        let acc = stop;
+        for (let i = overlays.length - 1; i >= 0; i--) acc = compositeColorOver(overlays[i], acc);
+        return acc;
+      });
+    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
+    if (bg && bg.a > 0.1) {
+      // An opaque surface above the gradient means the gradient never shows
+      // through here; resolveBackground would have returned it, so reaching
+      // this is defensive — bail rather than measure the wrong layer.
+      if (bg.a >= 0.99) return null;
+      overlays.push(bg);
+    }
     current = current.parentElement;
   }
   return null;
@@ -1986,6 +2109,10 @@ function checkElementColorsDOM(el) {
   const rect = el.getBoundingClientRect();
   if (rect.width < 10 || rect.height < 10) return [];
   const style = getComputedStyle(el);
+  // Invisible at rest: hidden scene variants (opacity-0 carousels, swap
+  // decks) are not user-visible, and measuring their inherited colors against
+  // whatever surface happens to sit behind the stack is noise, not audit.
+  if (style.visibility === 'hidden' || effectiveOpacityDOM(el) <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
   let effectiveBg = resolveBackground(el);
@@ -2798,11 +2925,13 @@ function checkElementGlowDOM(el) {
   // If resolveBackground returns null (gradient), try to infer from the gradient colors
   let parentBg = el.parentElement ? resolveBackground(el.parentElement) : resolveBackground(el);
   if (!parentBg) {
-    // Gradient background — sample its colors to determine if it's dark
+    // Gradient background — sample its colors to determine if it's dark.
+    // Modern-syntax parsing matters here: body-level gradients now reach this
+    // fallback in browser mode, and their stops usually serialize as oklch.
     let cur = el.parentElement;
     while (cur && cur.nodeType === 1) {
       const bgImage = getComputedStyle(cur).backgroundImage || '';
-      const gradColors = parseGradientColors(bgImage);
+      const gradColors = parseGradientColorsModern(bgImage);
       if (gradColors.length > 0) {
         // Average the gradient colors
         const avg = { r: 0, g: 0, b: 0 };
@@ -3644,6 +3773,14 @@ function checkElementBorders(tag, style, overrides, resolvedRadius, el = null) {
 }
 
 function checkElementColors(el, style, tag, window, customPropMap, hasAnchorInheritRule) {
+  // Invisible at rest, static twin of the browser walk's skip: opacity does
+  // not inherit, so walk ancestors multiplying declared opacity down.
+  if (style.visibility === 'hidden') return [];
+  let effOpacity = 1;
+  for (let cur = el; cur && cur.nodeType === 1 && effOpacity > 0.02; cur = cur.parentElement) {
+    effOpacity *= parseFloat(window.getComputedStyle(cur).opacity || '1');
+  }
+  if (effOpacity <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
 
@@ -4802,6 +4939,11 @@ function isRenderedForBrowserRule(el) {
 function checkElementTextOverflowDOM(el) {
   const tag = el.tagName.toLowerCase();
   if (TEXT_OVERFLOW_SKIP_TAGS.has(tag)) return [];
+  // scrollWidth/clientWidth are CSS box-model metrics; on SVG content Chrome
+  // returns arbitrary non-zero values for both (a <text> reported 78/48 while
+  // its rendered length sat comfortably inside its box), so the delta is
+  // noise, not overflow. SVG clips to its own viewport anyway.
+  if (el.namespaceURI === 'http://www.w3.org/2000/svg') return [];
   if (!isRenderedForBrowserRule(el)) return [];
   // Only the element that actually owns overflowing text — not its ancestors,
   // which inherit a wider scrollWidth from the spilling descendant.
@@ -5186,6 +5328,22 @@ function isPaintedForOcclusion(el) {
 // path is pure geometry and runs anywhere on the page.
 const OCCLUSION_TEXT_SKIP_TAGS = new Set(['script', 'style', 'noscript', 'template', 'title']);
 
+// An element whose effective opacity multiplies out to ~0 paints nothing at
+// rest: it is not user-visible, so visual findings on it (contrast, occlusion)
+// measure a state nobody sees. Browser-only — the walk needs live computed
+// styles. Cycling scenes that fade such elements in later are the screenshot
+// subsystem's territory, not the analytic walk's.
+function effectiveOpacityDOM(el) {
+  let o = 1;
+  // Walk all the way through body and html: `body { opacity: 0 }` page-fade
+  // wrappers hide every descendant just as thoroughly as a local wrapper.
+  for (let cur = el; cur && cur.nodeType === 1; cur = cur.parentElement) {
+    o *= parseFloat(getComputedStyle(cur).opacity || '1');
+    if (o <= 0.02) return 0;
+  }
+  return o;
+}
+
 function checkTextOcclusionDOM() {
   const findings = [];
   const seenVictims = new Set();
@@ -5213,6 +5371,11 @@ function checkTextOcclusionDOM() {
     }
     return false;
   };
+  // The classic occluder shape this rules out is an opacity-0 interaction
+  // layer — a range scrubber stretched over a before/after comparison — which
+  // elementFromPoint still returns and whose UA background-color otherwise
+  // reads as an opaque box.
+  const effectiveOpacity = effectiveOpacityDOM;
 
   // Collect renderable text owners in / near the first viewport for the
   // elementFromPoint probe. SVG <text> counts too.
@@ -5225,6 +5388,7 @@ function checkTextOcclusionDOM() {
     const text = inSvg ? (el.textContent || '').trim() : elementDirectText(el);
     if (text.length < 2) continue;
     if (!isPaintedForOcclusion(el)) continue;
+    if (effectiveOpacity(el) <= 0.02) continue;
     let rect; try { rect = el.getBoundingClientRect(); } catch { continue; }
     if (rect.width < 6 || rect.height < 6) continue;
     // Viewport-bound probe: keep text whose box overlaps the live viewport.
@@ -5258,6 +5422,7 @@ function checkTextOcclusionDOM() {
         if (top === el || el.contains(top) || top.contains(el)) continue;
         const topCs = getComputedStyle(top);
         if (isFloated(topCs) || isMarqueeish(top, topCs) || isPinnedOverlay(top)) continue;
+        if (effectiveOpacity(top) <= 0.02) continue;
         const topTag = top.tagName.toLowerCase();
         // Text sitting under a raw image/video is contrast territory (deduped
         // against the pixel low-contrast rule); leave those alone here.
@@ -5468,6 +5633,7 @@ export {
   CSS_NAMED_COLORS,
   checkBorders,
   isEmojiOnlyText,
+  scopedIgnoreActive,
   checkColors,
   checkHoverContrast,
   checkElementHoverContrast,

.claude/skills/impeccable/scripts/hook-before-edit.mjs

@@ -16,11 +16,15 @@ import path from 'node:path';
 
 import {
   ALLOWED_EXTS,
+  DEFAULT_CONFIG,
   EDIT_COUNT_THRESHOLD,
   GENERATED_PATH,
   SENSITIVE_PATH,
-  appendDesignSystemNote,
+  appendDesignSystemNoteOnce,
+  commitFooterShown,
+  designNoteReserve,
   designSystemOptions,
+  footerModeForSession,
   filterFindings,
   isNativePlatform,
   isScanTargetInsideProject,
@@ -345,13 +349,32 @@ async function detectProposedHtml(detector, content, filePath, scanOptions) {
   }
 }
 
-function cursorBlockMessage(findings, filePath, config, cwd) {
-  const rendered = renderTemplate(findings, filePath, config, { cwd });
-  const blocked = rendered.replace(
+// Cursor caps deny messages around 4000 chars. The cap feeds through the
+// renderer's clamp, which preserves the policy footer, rather than tail-
+// slicing the rendered text, which cut the footer off any message the
+// default 8000-char budget let past 4000.
+const CURSOR_DENY_LIMIT = 4000;
+const BLOCK_PREFIX = 'Impeccable design hook blocked this write before it landed. ';
+
+function cursorBlockMessage(findings, filePath, config, cwd, footerMode, reserveChars) {
+  const limits = config?.limits || DEFAULT_CONFIG.limits;
+  // Charge the prefix via reserveChars, not by subtracting from maxChars:
+  // renderTemplate's 500-char floor re-raises any maxChars pushed below it,
+  // un-charging a prefix subtracted from maxChars (Greptile P1 on PR #508).
+  // reserveChars comes off after the floor, so the prefix is charged at every
+  // config tier and the final prefixed message plus a pending staleness note
+  // fits the binding limit. Default-config output is byte-identical.
+  const budget = Math.min(
+    limits.maxChars || DEFAULT_CONFIG.limits.maxChars,
+    CURSOR_DENY_LIMIT,
+  );
+  const rendered = renderTemplate(findings, filePath,
+    { ...config, limits: { ...limits, maxChars: budget } },
+    { cwd, footer: footerMode, reserveChars: (reserveChars || 0) + BLOCK_PREFIX.length });
+  return rendered.replace(
     '[impeccable@1] Design hook findings requiring review',
-    '[impeccable@1] Impeccable design hook blocked this write before it landed. Design hook findings requiring review',
+    `[impeccable@1] ${BLOCK_PREFIX}Design hook findings requiring review`,
   );
-  return blocked.length > 4000 ? `${blocked.slice(0, 3984)}\n...(truncated)` : blocked;
 }
 
 function findingSignature(findings) {
@@ -468,9 +491,16 @@ async function main() {
     });
   }
 
-  const message = appendDesignSystemNote(cursorBlockMessage(filtered, filePath, config, cwd), scanOptions);
   const sessionId = event.session_id || event.conversation_id || 'unknown';
   const cache = readCache(cwd);
+  // Repeated denials for the same session repeat the findings, not the
+  // policy: the full footer emits once per session, the short form after.
+  const footerMode = footerModeForSession(cache, sessionId);
+  const message = appendDesignSystemNoteOnce(
+    cursorBlockMessage(filtered, filePath, config, cwd, footerMode, designNoteReserve(scanOptions, cache, sessionId)),
+    scanOptions, cache, sessionId, config,
+  );
+  commitFooterShown(cache, sessionId, message);
   const denial = bumpCursorDenial(cache, sessionId, filePath, filtered);
   persistCache(cwd, cache);
   if (denial.count > EDIT_COUNT_THRESHOLD) {

.claude/skills/impeccable/scripts/hook-lib.mjs

@@ -22,6 +22,9 @@
  *   dedupeAgainstCache(findings, cache, sessionId, filePath)
  *   renderTemplate(findings, filePath, config, opts)
  *   renderCleanAck(filePath, opts) / renderPendingAck(filePath, known, opts)
+ *   appendDesignSystemNote(text, scanOptions) / appendDesignSystemNoteOnce(text, scanOptions, cache, sessionId, config)
+ *   designNoteReserve(scanOptions, cache, sessionId)
+ *   footerModeForSession(cache, sessionId) / commitFooterShown(cache, sessionId, text)
  *   shouldEmitAckForFile(filePath, config?)
  *   writeAuditLog(env, entry)
  *   loadDetector() -> Promise<{ detectText, detectHtml }>
@@ -970,7 +973,13 @@ export function renderTemplate(findings, filePath, config, opts = {}) {
   if (!Array.isArray(findings) || findings.length === 0) return '';
   const limits = config?.limits || DEFAULT_CONFIG.limits;
   const cap = Math.max(1, limits.maxFindings || DEFAULT_CONFIG.limits.maxFindings);
-  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars);
+  // reserveChars holds back room for a note the caller appends after render
+  // (the DESIGN.md staleness note), so the final payload stays inside the
+  // configured budget. It comes off after the 500-char floor, so at floor
+  // configs the note keeps guaranteed delivery room; the clamp budget can
+  // therefore sit below 500, which clampLastLine's footer-preserving
+  // fallback handles (Bugbot on PR #508).
+  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars) - (opts.reserveChars || 0);
 
   const cwd = opts.cwd || process.cwd();
   const display = relativize(filePath, cwd);
@@ -979,11 +988,12 @@ export function renderTemplate(findings, filePath, config, opts = {}) {
   const remaining = total - shown.length;
 
   const header = `${ENVELOPE_PREFIX} Design hook findings requiring review in ${display} (${total} issue(s)):`;
-  const lines = shown.map((f) => formatFindingLine(f));
+  const seenRules = new Set();
+  const lines = shown.map((f) => formatDedupedFindingLine(f, seenRules));
   const more = remaining > 0
     ? `... and ${remaining} more (see ${IMPECCABLE_COMMAND} audit).`
     : null;
-  const footer = directiveFooter(display);
+  const footer = directiveFooter({ mode: opts.footer });
 
   const blocks = [header, ...lines];
   if (more) blocks.push(more);
@@ -1007,12 +1017,15 @@ function renderGroupedTemplate(groups, config, opts = {}) {
 
   const limits = config?.limits || DEFAULT_CONFIG.limits;
   const cap = Math.max(1, limits.maxFindings || DEFAULT_CONFIG.limits.maxFindings);
-  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars);
+  const maxChars = Math.max(500, limits.maxChars || DEFAULT_CONFIG.limits.maxChars) - (opts.reserveChars || 0);
   const cwd = opts.cwd || process.cwd();
   const total = realGroups.reduce((sum, group) => sum + group.findings.length, 0);
   const header = `${ENVELOPE_PREFIX} Design hook findings requiring review across ${realGroups.length} files (${total} issue(s)):`;
   const lines = [];
   let shownCount = 0;
+  // One seen-set across all groups: a rule already described under one file
+  // is not re-described under the next.
+  const seenRules = new Set();
 
   for (const group of realGroups) {
     const display = relativize(group.filePath, cwd);
@@ -1020,7 +1033,7 @@ function renderGroupedTemplate(groups, config, opts = {}) {
     const remainingCap = Math.max(0, cap - shownCount);
     const shown = group.findings.slice(0, remainingCap);
     for (const finding of shown) {
-      lines.push(formatFindingLine(finding));
+      lines.push(formatDedupedFindingLine(finding, seenRules));
     }
     shownCount += shown.length;
     const hidden = group.findings.length - shown.length;
@@ -1029,7 +1042,7 @@ function renderGroupedTemplate(groups, config, opts = {}) {
     }
   }
 
-  const footer = directiveFooter('the affected files', { grouped: true });
+  const footer = directiveFooter({ mode: opts.footer });
   let text = [header, ...lines, '', footer].join('\n');
   if (text.length > maxChars) {
     text = clampGroupedToBudget(header, lines, footer, maxChars);
@@ -1037,82 +1050,149 @@ function renderGroupedTemplate(groups, config, opts = {}) {
   return text;
 }
 
+// The clamp contract, shared by both budget functions: the footer is policy,
+// not detail, so it survives every clamp. Try the requested footer first;
+// when it cannot fit even after dropping finding lines, retry with the short
+// policy rather than sacrifice findings that fit beside it. A result that
+// dropped every finding line (a grouped render can fit a bare file header)
+// does not count as a fit: findings are why the emission exists.
+const isFindingLine = (line) => line.startsWith('- ');
+
+function footerFallbacks(footer) {
+  const short = directiveFooter({ mode: 'short' });
+  return footer === short ? [footer] : [footer, short];
+}
+
 function clampGroupedToBudget(header, lines, footer, maxChars) {
-  const assemble = (linesArr, omitted) => [
+  const assemble = (linesArr, omitted, footerText) => [
     header,
     ...linesArr,
     ...(omitted ? [`... and more (see ${IMPECCABLE_COMMAND} audit).`] : []),
     '',
-    footer,
+    footerText,
   ].join('\n');
 
-  let working = lines.slice();
-  let omitted = false;
-  let assembled = assemble(working, omitted);
-  while (assembled.length > maxChars && working.length > 1) {
-    working.pop();
-    omitted = true;
-    assembled = assemble(working, omitted);
+  for (const footerText of footerFallbacks(footer)) {
+    let working = lines.slice();
+    let omitted = false;
+    let assembled = assemble(working, omitted, footerText);
+    while (assembled.length > maxChars && working.length > 1) {
+      working.pop();
+      omitted = true;
+      assembled = assemble(working, omitted, footerText);
+    }
+    if (assembled.length <= maxChars && working.some(isFindingLine)) return assembled;
   }
-  if (assembled.length > maxChars) {
-    assembled = `${assembled.slice(0, maxChars - 1)}…`;
-  }
-  return assembled;
+  return clampLastLine((linesArr, footerText) => assemble(linesArr, true, footerText),
+    lines.find(isFindingLine) || lines[0], maxChars);
 }
 
 function clampToBudget(header, lines, more, footer, maxChars) {
-  const assemble = (linesArr, moreText) => {
+  const assemble = (linesArr, moreText, footerText) => {
     const blocks = [header, ...linesArr];
     if (moreText) blocks.push(moreText);
     blocks.push('');
-    blocks.push(footer);
+    blocks.push(footerText);
     return blocks.join('\n');
   };
 
-  let working = lines.slice();
-  let moreText = more;
-  let assembled = assemble(working, moreText);
-  while (assembled.length > maxChars && working.length > 1) {
-    working.pop();
-    moreText = `... and more (see ${IMPECCABLE_COMMAND} audit).`;
-    assembled = assemble(working, moreText);
+  let lastMore = more;
+  for (const footerText of footerFallbacks(footer)) {
+    let working = lines.slice();
+    let moreText = more;
+    let assembled = assemble(working, moreText, footerText);
+    while (assembled.length > maxChars && working.length > 1) {
+      working.pop();
+      moreText = `... and more (see ${IMPECCABLE_COMMAND} audit).`;
+      assembled = assemble(working, moreText, footerText);
+    }
+    lastMore = moreText;
+    if (assembled.length <= maxChars) return assembled;
   }
-  if (assembled.length > maxChars) {
-    assembled = `${assembled.slice(0, maxChars - 1)}…`;
-  }
-  return assembled;
+  return clampLastLine((linesArr, footerText) => assemble(linesArr, lastMore, footerText),
+    lines.find(isFindingLine) || lines[0], maxChars);
 }
 
-function formatFindingLine(f) {
+// Last resort with one finding line left: the short policy gets the budget
+// first, the line is clipped to what remains. The pre-fix tail-slice cut
+// whatever happened to be last, which was always the footer.
+function clampLastLine(build, line, maxChars) {
+  const footerText = directiveFooter({ mode: 'short' });
+  const bare = build([], footerText);
+  // +1 for the newline the line itself brings when it joins the blocks.
+  const room = maxChars - bare.length - 1;
+  if (room >= 24) {
+    const clipped = line.length > room ? `${line.slice(0, room - 1)}…` : line;
+    return build([clipped], footerText);
+  }
+  // No room for even a clipped finding line: the note reservation can pull
+  // the budget below the 500-char floor, and a deep file path can push the
+  // header past what remains beside the short policy (Bugbot on PR #508).
+  // Drop the line, and if the bare header + policy still overflow, clip the
+  // head. Never tail-slice: the footer sits at the end, so a tail slice is
+  // exactly the footer cut this renderer exists to prevent.
+  if (bare.length <= maxChars) return bare;
+  const head = bare.slice(0, Math.max(0, maxChars - footerText.length - 4));
+  return `${head}…\n\n${footerText}`;
+}
+
+// `compact` drops the registry description: within one emission the first
+// occurrence of a rule carries the full description and repeats keep only the
+// rule id, name, and their own ignore hint (values differ per line, so the
+// hint must survive the dedupe).
+function formatFindingLine(f, opts = {}) {
   const prefix = f.line && f.line > 0 ? `- L${f.line}` : '-';
-  const desc = (f.description || '').trim();
+  const desc = opts.compact ? '' : (f.description || '').trim();
   const name = (f.name || '').trim();
   // Description from the registry already ends in punctuation; join with a
   // single space. `name` may have a trailing period already, keep it clean.
   const nameSegment = name ? `${name.replace(/\.+\s*$/, '')}.` : '';
-  const ignoreCommand = formatFindingIgnoreCommand(f);
-  const ignoreSegment = ignoreCommand
-    ? ` If the user explicitly confirms this value is intentional: \`${ignoreCommand}\`.`
-    : '';
+  const ignoreHint = formatFindingIgnoreHint(f);
+  const ignoreSegment = ignoreHint ? ` If intentional: \`${ignoreHint}\`.` : '';
   return `${prefix} [${f.antipattern}] ${nameSegment} ${desc}${ignoreSegment}`.replace(/\s+/g, ' ').trim();
 }
 
-function formatFindingIgnoreCommand(finding) {
+// Dedupe applied in shown-line order, so the first rendered occurrence of a
+// rule always carries the description. The budget clamps pop lines from the
+// end, which can never orphan a compact repeat before its described first
+// occurrence.
+function formatDedupedFindingLine(finding, seenRules) {
+  const rule = normalizeIgnoreRule(finding?.antipattern);
+  const compact = rule ? seenRules.has(rule) : false;
+  if (rule) seenRules.add(rule);
+  return formatFindingLine(finding, { compact });
+}
+
+// The rule/value pair the footer's `hook-admin.mjs ignore-value` command
+// takes. Deliberately just the args: the executable prefix, the --reason
+// contract, and the disclosure rule live in the directive footer, stated once
+// instead of per line.
+function formatFindingIgnoreHint(finding) {
   if (!finding || typeof finding !== 'object') return '';
   const rule = normalizeIgnoreRule(finding.antipattern);
   if (!rule) return '';
   const normalizedValue = extractFindingIgnoreValue(finding);
   if (!normalizedValue) return '';
-  const value = extractFindingIgnoreValueRaw(finding);
-  const valueArg = quoteCommandArg(value);
-  const reason = quoteCommandArg(`User confirmed ${value} is intentional`);
-  return `${IMPECCABLE_COMMAND} hooks ignore-value ${rule} ${valueArg} --shared --reason ${reason}`;
+  const valueArg = quoteCommandArg(extractFindingIgnoreValueRaw(finding));
+  return `ignore-value ${rule} ${valueArg}`;
 }
 
 function quoteCommandArg(value) {
   const text = String(value || '').trim();
   if (/^[A-Za-z0-9._:-]+$/.test(text)) return text;
-  return `"${text.replace(/\\/g, '\\\\').replace(/"/g, '\\"')}"`;
+  // The suggestion is meant to be run on this same machine, so quote for its
+  // shell. POSIX /bin/sh still expands $(...), backticks, and ${} inside
+  // double quotes, and these values come from scanned file content (a
+  // font-family name) or a file path, so untrusted input must be
+  // single-quoted (issue #476). Windows cmd.exe performs no such command
+  // substitution, but it treats a single quote as a literal character rather
+  // than a grouping delimiter, so a value or path containing spaces has to
+  // stay double-quoted there (Greptile #533). Keep the pre-existing
+  // double-quote escaping on Windows so that path's behavior is unchanged.
+  if (process.platform === 'win32') {
+    return `"${text.replace(/\\/g, '\\\\').replace(/"/g, '\\"')}"`;
+  }
+  return `'${text.replace(/'/g, `'\\''`)}'`;
 }
 
 function relativize(filePath, cwd) {
@@ -1594,36 +1674,105 @@ export function designSystemOptions(config, detector, projectCwd) {
   }
 }
 
+const DESIGN_STALE_NOTE = `${ENVELOPE_PREFIX} DESIGN.md is newer than .impeccable/design.json. Run ${IMPECCABLE_COMMAND} document to refresh the design-system sidecar.`;
+
 export function appendDesignSystemNote(text, scanOptions) {
   if (!text || !scanOptions?.designSystem?.mdNewerThanJson) return text;
-  return `${text}\n\n${ENVELOPE_PREFIX} DESIGN.md is newer than .impeccable/design.json. Run ${IMPECCABLE_COMMAND} document to refresh the design-system sidecar.`;
+  return `${text}\n\n${DESIGN_STALE_NOTE}`;
 }
 
+// Session-scoped once-only gate for repeat-prone message parts. Returns true
+// the first time a flag is consumed in a session and false after, mirroring
+// the `cleanAcked` mechanic: the mtime skew (and the policy footer) do not
+// change between edits, so re-stating them on every emission spends context
+// to say nothing new. Callers must persist the cache for the flag to stick.
+function consumeSessionNoticeFlag(cache, sessionId, flag) {
+  const session = ensureSession(cache, sessionId);
+  if (session[flag]) return false;
+  session[flag] = true;
+  session.updatedAt = Date.now();
+  return true;
+}
+
+// Once-per-session variant of appendDesignSystemNote for the emission paths
+// that have cache access. The staleness note names standing project state,
+// not new information, so one mention per session is enough. The note is
+// appended after the renderer has clamped to the configured budget: render
+// paths reserve room for it via designNoteReserve, and the size check here
+// is the safety net for the ack paths, deferring (without consuming the
+// flag) to a later emission rather than busting maxChars.
+export function appendDesignSystemNoteOnce(text, scanOptions, cache, sessionId, config) {
+  if (!text || !scanOptions?.designSystem?.mdNewerThanJson) return text;
+  const maxChars = Math.max(500, config?.limits?.maxChars || DEFAULT_CONFIG.limits.maxChars);
+  if (text.length + DESIGN_STALE_NOTE.length + 2 > maxChars) return text;
+  if (!consumeSessionNoticeFlag(cache, sessionId, 'designNoteShown')) return text;
+  return appendDesignSystemNote(text, scanOptions);
+}
+
+// Render-time reservation for the note above: how many characters the
+// renderer must hold back so a pending staleness note still fits inside the
+// configured budget. Zero once the session has seen the note. Without the
+// reservation, a session whose every emission fills the budget would defer
+// the note forever.
+export function designNoteReserve(scanOptions, cache, sessionId) {
+  if (!scanOptions?.designSystem?.mdNewerThanJson) return 0;
+  if (ensureSession(cache, sessionId).designNoteShown) return 0;
+  return DESIGN_STALE_NOTE.length + 2;
+}
+
+// Full directive footer once per session, the short reminder after. Fresh
+// emissions and Cursor denials share the session flag (`footerShown`), so a
+// session pays the full policy exactly once however it first fires. The mode
+// is a peek: the clamp can downgrade a requested full footer under a tight
+// budget, so the flag commits only when the complete full policy actually
+// reached the output. Matching the whole footer text (not a sentinel) keeps
+// the flag honest against any truncation that spares the opening words.
+export function footerModeForSession(cache, sessionId) {
+  return ensureSession(cache, sessionId).footerShown ? 'short' : 'full';
+}
+
+export function commitFooterShown(cache, sessionId, text) {
+  if (!text || !text.includes(directiveFooter())) return;
+  const session = ensureSession(cache, sessionId);
+  if (session.footerShown) return;
+  session.footerShown = true;
+  session.updatedAt = Date.now();
+}
+
+const HOOK_ADMIN_COMMAND = `node ${quoteCommandArg(path.join(__dirname, 'hook-admin.mjs'))}`;
+
 // The directive footer is the part of the hook output that steers model
-// behavior. Three intentional moves:
-//   1. **Imperative, not advisory.** "Handle these..." beats "Consider
-//      revising..." which the model treats as a soft suggestion it can
-//      override when the user asked for any kind of throwaway / demo UI.
-//   2. **Explicit judgment clause.** Without it, the model will try to
-//      "fix" intentional motion, bad fixtures, anti-pattern examples in
-//      docs, or test cases. Naming the judgment inline beats hoping the
-//      model infers it from context.
-//   3. **Acknowledgement instruction.** Hook output is injected as
-//      developer-role context, not a chat turn, so the user never sees the
-//      raw envelope. Asking the model to surface the resolution in its
-//      reply is the cheapest way to make the feedback loop visible.
-function directiveFooter(display, opts = {}) {
-  // Offer the rule-scoped-to-file form first. `ignore-file` silences every rule
-  // for the path forever, which is far more than one noisy rule on a real UI
-  // surface justifies, and it was previously the only option named here.
-  const target = opts.grouped ? '<path>' : quoteCommandArg(display);
-  const fileIgnoreGuidance = `run \`${IMPECCABLE_COMMAND} hooks ignore-value <id> "*" --file ${target}\` to scope just that rule to the file, or \`${IMPECCABLE_COMMAND} hooks ignore-file ${target}\` only when the whole file is out of scope for design review (a fixture, a generated artifact, a deliberate demo)`;
+// behavior. Intentional moves, in order:
+//   1. **Imperative, not advisory.** "Triage each finding..." beats
+//      "Consider revising...", which the model treats as a soft suggestion.
+//   2. **Positive triage branches.** Fix / suppress-and-disclose / ask. The
+//      suppress branch names the calibration examples (demo, fixture,
+//      documented bad design, user-confirmed choice) because the agent now
+//      acts on its own confidence and needs the bar stated.
+//   3. **Executable ignore path.** The old footer named only the slash
+//      command, which an agent reacting to hook output cannot run; the
+//      hook-admin.mjs invocation is runnable as-is and keeps agents out of
+//      hand-editing config.json.
+//   4. **Honest provenance.** The --reason is the audit trail; "user
+//      confirmed" appears only when the user actually did.
+//   5. **Acknowledgement instruction.** Hook output is injected as
+//      developer-role context, so the reply is where the user sees the
+//      resolution, including any ignore the agent persisted.
+//   6. **Once per session.** The full policy emits on the session's first
+//      fire; later emissions carry the one-line short form (mode 'short').
+function directiveFooter(opts = {}) {
+  if (opts.mode === 'short') {
+    // No command path here: the session's first emission already gave the
+    // runnable hook-admin.mjs invocation, and restating ~70 chars of absolute
+    // path on every repeat is the duplication this mode exists to cut.
+    return 'Triage per the session policy: fix real problems; persist confident false-positive or sanctioned-exception ignores via `hook-admin.mjs ignore-value` and disclose them in your reply; unsure, ask in one line.';
+  }
   return [
-    'Handle these before finalizing: fix findings that are real design problems, or explicitly classify contextually intentional findings as false positives. Acknowledge what you changed or why you are leaving a finding unchanged.',
-    '',
-    'Use context judgment before editing. A finding is not automatically a defect; literal or domain-appropriate motion, intentional demos or fixtures, documentation of bad design, and user-confirmed choices can be valid as-is.',
-    '',
-    `Do not change intentional design just to satisfy the hook, and do not silence a real finding with an inline ignore comment to skip fixing it. Suppress a finding only after the user explicitly confirms it is intentional. Prefer a config ignore (one reviewable place, the commands below); reach for an inline \`impeccable-disable <rule>\` comment only when the waiver must travel with a file that leaves the repo, such as an exported or standalone document. Prefer the narrowest persisted exception: run the exact \`${IMPECCABLE_COMMAND} hooks ignore-value ... --shared\` command shown next to a value-specific finding. For \`overused-font\`, use \`ignore-value\` for a specific font and use \`${IMPECCABLE_COMMAND} hooks ignore-rule overused-font --all-values\` only when the user asks to ignore overused fonts generally. For a finding whose line shows no exact ignore-value command, such as \`side-tab\`, ${fileIgnoreGuidance}; use \`${IMPECCABLE_COMMAND} hooks ignore-rule <id>\` only when the user asks to suppress the whole non-value-specific rule. Run ${IMPECCABLE_COMMAND} audit for the full pass.`,
+    'Triage each finding, then state in your reply what you fixed, what you suppressed, and what you left standing:',
+    '- Real design problem: fix it. Keep intentional design as designed.',
+    `- Confident false positive or sanctioned exception (an intentional demo or fixture, documentation of bad design, literal or domain-appropriate motion, a choice the user confirmed): persist the narrowest ignore yourself and disclose it. Run \`${HOOK_ADMIN_COMMAND} ignore-value <rule> "<value>" --reason "<who decided: evidence>"\` with the pair shown on the finding line, or value "*" plus \`--file <path>\` when the line shows none. Write "user confirmed" in a reason only when the user did.`,
+    '- Unsure: leave it as is and ask the user in one line.',
+    `Self-serve ends at ignore-value: \`ignore-file\` and \`ignore-rule\` need the user's explicit approval, and never add an ignore to push a blocked write through. Full suppression ladder: ${IMPECCABLE_COMMAND} hooks.`,
   ].join('\n');
 }
 
@@ -1845,20 +1994,23 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       }
     }
 
-    // Persist only when the write is earned: fresh findings justify creating
-    // `.impeccable/` (dedup and suppression need it), deferred findings do
-    // too (the Stop deep pass needs the touched-file list to surface them),
-    // and an already-present `.impeccable/` dir marks a project that opted
-    // in. A non-UI edit, or a clean UI edit in a project with no Impeccable
-    // footprint, must be a no-op on disk (issues #344, #305).
-    if (freshGroups.length > 0 || deferredTotal > 0
-      || (cacheDirty && fs.existsSync(path.join(projectCwd, '.impeccable')))) {
-      persistCache(projectCwd, cache);
-    }
-
+    // The session notice flags mutate the cache, so they must settle before
+    // the persist that makes them stick across events.
     if (freshGroups.length > 0) {
       const firstGroup = freshGroups[0];
-      const text = appendDesignSystemNote(renderGroupedTemplate(freshGroups, config, { cwd: projectCwd }), scanOptions);
+      const footerMode = footerModeForSession(cache, sessionId);
+      const text = appendDesignSystemNoteOnce(
+        renderGroupedTemplate(freshGroups, config, {
+          cwd: projectCwd,
+          footer: footerMode,
+          reserveChars: designNoteReserve(scanOptions, cache, sessionId),
+        }),
+        scanOptions, cache, sessionId, config,
+      );
+      commitFooterShown(cache, sessionId, text);
+      // Fresh findings always earn the cache write, including creating
+      // `.impeccable/`: dedup, suppression, and the notice flags need it.
+      persistCache(projectCwd, cache);
       const allFindings = freshGroups.flatMap((group) => group.findings);
       return {
         exitCode: 0,
@@ -1881,6 +2033,33 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       };
     }
 
+    // Resolve the ack emission before the persist below: appendDesignSystem-
+    // NoteOnce consumes a session flag, and the flag only sticks when the
+    // write happens after it. Quiet mode emits nothing, so it consumes
+    // nothing. The clean arm mirrors the branch order further down: pending
+    // outranks suppression, suppression outranks clean.
+    let ack = null;
+    if (!quietMode && pendingWinner && shouldEmitAckForFile(pendingWinner.filePath, config)) {
+      ack = {
+        kind: 'pending',
+        text: appendDesignSystemNoteOnce(renderPendingAck(pendingWinner.filePath, pendingWinner.known, { cwd: projectCwd }), scanOptions, cache, sessionId, config),
+      };
+    } else if (!quietMode && !suppressionWinner && cleanWinner && !cleanAckDeduped && shouldEmitAckForFile(cleanWinner.filePath, config)) {
+      ack = {
+        kind: 'clean',
+        text: appendDesignSystemNoteOnce(renderCleanAck(cleanWinner.filePath, { cwd: projectCwd }), scanOptions, cache, sessionId, config),
+      };
+    }
+
+    // Persist only when the write is earned: deferred findings need the
+    // touched-file list for the Stop deep pass, and an already-present
+    // `.impeccable/` dir marks a project that opted in. A non-UI edit, or a
+    // clean UI edit in a project with no Impeccable footprint, must be a
+    // no-op on disk (issues #344, #305).
+    if (deferredTotal > 0 || (cacheDirty && fs.existsSync(path.join(projectCwd, '.impeccable')))) {
+      persistCache(projectCwd, cache);
+    }
+
     if (detectorThrewAny && !pendingWinner && !cleanWinner) {
       return result({ emitted: false, error: 'detector-threw', durationMs: Date.now() - started });
     }
@@ -1889,8 +2068,8 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       return result({ emitted: false, quiet: true, durationMs: Date.now() - started });
     }
 
-    if (pendingWinner && shouldEmitAckForFile(pendingWinner.filePath, config)) {
-      const text = appendDesignSystemNote(renderPendingAck(pendingWinner.filePath, pendingWinner.known, { cwd: projectCwd }), scanOptions);
+    if (ack?.kind === 'pending') {
+      const text = ack.text;
       return {
         exitCode: 0,
         stdout: payload(text, 'PostToolUse', harness),
@@ -1923,8 +2102,8 @@ export async function runHook({ stdinJson, env = {}, cwd = process.cwd(), now =
       };
     }
 
-    if (cleanWinner && !cleanAckDeduped && shouldEmitAckForFile(cleanWinner.filePath, config)) {
-      const text = appendDesignSystemNote(renderCleanAck(cleanWinner.filePath, { cwd: projectCwd }), scanOptions);
+    if (ack?.kind === 'clean') {
+      const text = ack.text;
       return {
         exitCode: 0,
         stdout: payload(text, 'PostToolUse', harness),
@@ -2108,11 +2287,22 @@ export async function runStopHook({ stdinJson, env = {}, cwd = process.cwd(), no
       return result({ emitted: false, skipped: 'stop-clean', durationMs: Date.now() - started });
     }
 
-    // Fresh findings earn the cache write so the next Stop fire is silent
-    // unless new issues appear.
-    persistCache(projectCwd, cache);
+    // A per-edit fire earlier in this session already consumed the footer
+    // flag, so the Stop wall of text carries the one-line short footer.
+    const footerMode = footerModeForSession(cache, sessionId);
+    const text = appendDesignSystemNoteOnce(
+      renderGroupedTemplate(freshGroups, config, {
+        cwd: projectCwd,
+        footer: footerMode,
+        reserveChars: designNoteReserve(scanOptions, cache, sessionId),
+      }),
+      scanOptions, cache, sessionId, config,
+    );
+    commitFooterShown(cache, sessionId, text);
 
-    const text = appendDesignSystemNote(renderGroupedTemplate(freshGroups, config, { cwd: projectCwd }), scanOptions);
+    // Fresh findings earn the cache write so the next Stop fire is silent
+    // unless new issues appear; the notice flags ride along.
+    persistCache(projectCwd, cache);
     return {
       exitCode: 0,
       stdout: payload(text, 'Stop', harness),

.claude/skills/impeccable/scripts/lib/concept-catalog.mjs

@@ -109,6 +109,18 @@ export function validateConceptEntry(concept, { existingForms = new Map(), axes
     || concept.tags.some(tag => typeof tag !== 'string' || !tag.trim())) {
     errors.push(`concept ${id} must have exactly three structural tags`);
   }
+  // The slop this world in particular is at risk of. Optional, because 541
+  // entries predate it and none of them are wrong for lacking it. A world built
+  // from posters is at risk of shouting and one built from instruments is at
+  // risk of dead greys; a global detector cannot know which, and the author can.
+  if (concept?.avoid !== undefined) {
+    if (!Array.isArray(concept.avoid)
+      || concept.avoid.length < 2
+      || concept.avoid.length > 3
+      || concept.avoid.some(item => typeof item !== 'string' || item.trim().length < 12 || item.trim().length > 160)) {
+      errors.push(`concept ${id} avoid must be two or three negations of 12–160 characters`);
+    }
+  }
   if (!Array.isArray(concept?.system)
     || concept.system.length !== SYSTEM_PREFIXES.length
     || concept.system.some(rule => typeof rule !== 'string' || rule.trim().length < 12 || rule.trim().length > 180)) {

.claude/skills/impeccable/scripts/lib/impeccable-config.mjs

@@ -206,10 +206,10 @@ function parseIgnoreColor(value) {
   if (rgb) {
     const parts = splitColorArgs(rgb[1]);
     if (parts.length < 3 || parts.length > 4) return null;
-    const r = parseRgbChannel(parts[0]);
-    const g = parseRgbChannel(parts[1]);
-    const b = parseRgbChannel(parts[2]);
-    const a = parts[3] === undefined ? 1 : parseAlphaChannel(parts[3]);
+    const r = parseColorChannel(parts[0], COLOR_CHANNEL_FORMATS.rgb);
+    const g = parseColorChannel(parts[1], COLOR_CHANNEL_FORMATS.rgb);
+    const b = parseColorChannel(parts[2], COLOR_CHANNEL_FORMATS.rgb);
+    const a = parts[3] === undefined ? 1 : parseColorChannel(parts[3], COLOR_CHANNEL_FORMATS.alpha);
     if ([r, g, b, a].some((v) => v === null)) return null;
     return { r, g, b, a };
   }
@@ -218,10 +218,10 @@ function parseIgnoreColor(value) {
   if (hsl) {
     const parts = splitColorArgs(hsl[1]);
     if (parts.length < 3 || parts.length > 4) return null;
-    const h = parseHueChannel(parts[0]);
-    const s = parsePercentChannel(parts[1]);
-    const l = parsePercentChannel(parts[2]);
-    const a = parts[3] === undefined ? 1 : parseAlphaChannel(parts[3]);
+    const h = parseColorChannel(parts[0], COLOR_CHANNEL_FORMATS.hue);
+    const s = parseColorChannel(parts[1], COLOR_CHANNEL_FORMATS.percent);
+    const l = parseColorChannel(parts[2], COLOR_CHANNEL_FORMATS.percent);
+    const a = parts[3] === undefined ? 1 : parseColorChannel(parts[3], COLOR_CHANNEL_FORMATS.alpha);
     if ([h, s, l, a].some((v) => v === null)) return null;
     return hslToRgb(h, s, l, a);
   }
@@ -230,18 +230,13 @@ function parseIgnoreColor(value) {
 }
 
 function parseHexIgnoreColor(hex) {
-  if (hex.length === 3 || hex.length === 4) {
-    const r = parseInt(hex[0] + hex[0], 16);
-    const g = parseInt(hex[1] + hex[1], 16);
-    const b = parseInt(hex[2] + hex[2], 16);
-    const a = hex.length === 4 ? parseInt(hex[3] + hex[3], 16) / 255 : 1;
-    return { r, g, b, a };
-  }
-  const r = parseInt(hex.slice(0, 2), 16);
-  const g = parseInt(hex.slice(2, 4), 16);
-  const b = parseInt(hex.slice(4, 6), 16);
-  const a = hex.length === 8 ? parseInt(hex.slice(6, 8), 16) / 255 : 1;
-  return { r, g, b, a };
+  const expanded = hex.length <= 4
+    ? [...hex].map((digit) => digit.repeat(2)).join('')
+    : hex;
+  const [r, g, b, alpha = 255] = expanded
+    .match(/../g)
+    .map((channel) => Number.parseInt(channel, 16));
+  return { r, g, b, a: alpha / 255 };
 }
 
 function splitColorArgs(body) {
@@ -259,47 +254,34 @@ function splitColorArgs(body) {
   return text.replace(/\s*\/\s*/g, ' / ').split(/\s+/).filter((part) => part && part !== '/');
 }
 
-function parseRgbChannel(raw) {
-  const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)(%)?$/);
-  if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  const scaled = match[2] ? value * 2.55 : value;
-  if (scaled < 0 || scaled > 255) return null;
-  return Math.round(scaled);
-}
+const CSS_NUMBER_RE = /^(-?\d*\.?\d+)(%|deg|rad|turn|grad)?$/;
+const identity = (value) => value;
+const COLOR_CHANNEL_FORMATS = {
+  rgb: { units: { '': identity, '%': (value) => value * 2.55 }, min: 0, max: 255, round: true },
+  alpha: { units: { '': identity, '%': (value) => value / 100 }, min: 0, max: 1 },
+  hue: {
+    units: {
+      '': identity,
+      deg: identity,
+      rad: (value) => value * (180 / Math.PI),
+      turn: (value) => value * 360,
+      grad: (value) => value * 0.9,
+    },
+  },
+  percent: { units: { '%': (value) => value / 100 }, min: 0, max: 1 },
+};
 
-function parseAlphaChannel(raw) {
+function parseColorChannel(raw, { units, min = -Infinity, max = Infinity, round = false }) {
   const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)(%)?$/);
+  const match = text.match(CSS_NUMBER_RE);
   if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  const alpha = match[2] ? value / 100 : value;
-  return alpha >= 0 && alpha <= 1 ? alpha : null;
-}
-
-function parseHueChannel(raw) {
-  const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)(deg|rad|turn|grad)?$/);
-  if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  const unit = match[2] || 'deg';
-  if (unit === 'turn') return value * 360;
-  if (unit === 'rad') return value * (180 / Math.PI);
-  if (unit === 'grad') return value * 0.9;
-  return value;
-}
-
-function parsePercentChannel(raw) {
-  const text = String(raw || '').trim();
-  const match = text.match(/^(-?\d*\.?\d+)%$/);
-  if (!match) return null;
-  const value = Number.parseFloat(match[1]);
-  if (!Number.isFinite(value)) return null;
-  return value >= 0 && value <= 100 ? value / 100 : null;
+  const convert = units[match[2] || ''];
+  if (!convert) return null;
+  const number = Number.parseFloat(match[1]);
+  if (!Number.isFinite(number)) return null;
+  const value = convert(number);
+  if (value < min || value > max) return null;
+  return round ? Math.round(value) : value;
 }
 
 function hslToRgb(hue, saturation, lightness, alpha) {

.claude/skills/impeccable/scripts/lib/is-generated.mjs

@@ -13,7 +13,7 @@
  *      within the first ~300 characters — catches non-git projects.
  */
 
-import { execSync } from 'node:child_process';
+import { execFileSync } from 'node:child_process';
 import fs from 'node:fs';
 import path from 'node:path';
 
@@ -41,7 +41,10 @@ export function isGeneratedFile(filePath, options = {}) {
 
 function isGitIgnored(absPath, cwd) {
   try {
-    execSync(`git check-ignore --quiet ${JSON.stringify(absPath)}`, {
+    // argv form, never a shell: this runs on every file the live-mode source
+    // walk reaches, so a hostile filename embedding $(...) or backticks must
+    // not be interpretable (issue #476). JSON.stringify is not shell quoting.
+    execFileSync('git', ['check-ignore', '--quiet', absPath], {
       cwd,
       stdio: 'ignore',
     });

.claude/skills/impeccable/scripts/lib/open-system-browser.mjs

@@ -0,0 +1,26 @@
+import { spawn } from 'node:child_process';
+
+export function browserOpenCommand(url, {
+  platform = process.platform,
+  comspec = process.env.ComSpec || process.env.COMSPEC || 'cmd.exe',
+} = {}) {
+  if (platform === 'darwin') return { command: 'open', args: [url] };
+  if (platform === 'win32') return { command: comspec, args: ['/c', 'start', '', url] };
+  return { command: 'xdg-open', args: [url] };
+}
+
+export function openSystemBrowser(url, {
+  platform = process.platform,
+  comspec = process.env.ComSpec || process.env.COMSPEC || 'cmd.exe',
+  spawnImpl = spawn,
+} = {}) {
+  const { command, args } = browserOpenCommand(url, { platform, comspec });
+  try {
+    const child = spawnImpl(command, args, { stdio: 'ignore', detached: true });
+    child.on('error', () => {});
+    child.unref();
+    return true;
+  } catch {
+    return false;
+  }
+}

.claude/skills/impeccable/scripts/lib/staleness-deep.mjs

@@ -244,7 +244,8 @@ const HOOK_MARKER = /skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs/;
 //   * bundle-relative: node ".agents/.../hook.mjs"
 //   * legacy unquoted: node .claude/.../hook.mjs
 //   * guarded (#399):  [ ! -f "PATH" ] || node "PATH"   (PATH twice, identical)
-//   * absolute:        node "/Users/.../hook.mjs"        (user-level installs)
+//   * absolute (#476): [ ! -f 'PATH' ] || node 'PATH'   (single-quoted since
+//                      the shell-injection fix; older installs double-quote)
 //   * github portable: node "$(git rev-parse --show-toplevel)/.../hook.mjs"
 // A quoted path wins; the guard's two occurrences are identical, so the first
 // quoted match is the path. Otherwise fall back to the whitespace/metachar-
@@ -255,6 +256,12 @@ function hookScriptTokenFrom(command) {
   if (!HOOK_MARKER.test(str)) return null;
   const quoted = str.match(/"([^"]*skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs)"/);
   if (quoted) return quoted[1];
+  // A path containing an apostrophe serializes as '\'' inside single quotes;
+  // no regex reassembles that, and the bare fallback would misread a fragment
+  // of it, so return null: the caller never asserts on a path it can't parse.
+  if (str.includes("'\\''")) return null;
+  const singleQuoted = str.match(/'([^']*skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs)'/);
+  if (singleQuoted) return singleQuoted[1];
   const bare = str.match(/([^\s"'|&;()]*skills\/impeccable\/scripts\/hook(?:-before-edit)?\.mjs)/);
   return bare ? bare[1] : null;
 }

.claude/skills/impeccable/scripts/live-browser.js

@@ -97,23 +97,20 @@
     return { value: c.value, label: c.label };
   });
 
-  const LIVE_CHROME_MOUNT_CONTRACT = ['root', 'transport', 'state', 'actions'];
-  const LIVE_UI_SURFACES = [
-    { key: 'global-bottom-bar', ids: [PREFIX + '-global-bar', PREFIX + '-global-bar-brand', PREFIX + '-pick-toggle', PREFIX + '-insert-toggle', PREFIX + '-detect-toggle', PREFIX + '-detect-badge', PREFIX + '-design-toggle', PREFIX + '-page-chat', PREFIX + '-page-chat-input', PREFIX + '-page-chat-voice', PREFIX + '-page-chat-send'] },
-    { key: 'pending-copy-edit-dock', ids: [PREFIX + '-pending-dock'] },
-    { key: 'element-selection-chrome', ids: [PREFIX + '-highlight', PREFIX + '-tooltip', PREFIX + '-bar', PREFIX + '-selection-pill', PREFIX + '-input', PREFIX + '-configure-voice', PREFIX + '-configure-bar-tooltip'] },
-    { key: 'action-picker', ids: [PREFIX + '-picker'] },
-    { key: 'edit-chrome', ids: [PREFIX + '-edit-badge'] },
-    { key: 'generating-row', ids: [PREFIX + '-bar', PREFIX + '-shader'] },
-    { key: 'variant-cycling-row', ids: [PREFIX + '-bar', PREFIX + '-params-panel'] },
-    { key: 'variant-params-panel', ids: [PREFIX + '-params-panel'] },
-    { key: 'saving-confirmed-rows', ids: [PREFIX + '-bar'] },
-    { key: 'insert-mode-chrome', ids: [PREFIX + '-insert-line', PREFIX + '-insert-placeholder', PREFIX + '-placeholder-resize', PREFIX + '-insert-input', PREFIX + '-insert-voice', PREFIX + '-insert-create', PREFIX + '-insert-create-tooltip'] },
-    { key: 'annotation-chrome', ids: [PREFIX + '-annot', PREFIX + '-annot-svg', PREFIX + '-annot-pins', PREFIX + '-annot-clear'] },
-    { key: 'design-system-panel', ids: [PREFIX + '-design-host'] },
-    { key: 'toasts-and-errors', ids: [PREFIX + '-toast', PREFIX + '-mount-error'] },
-    { key: 'css-isolation-boundary', ids: [PREFIX + '-root'] },
-  ];
+  // The Live chrome inventory (which surfaces exist, and the element ids each
+  // one owns) comes from the canonical source, skill/scripts/live/ui-surfaces.mjs,
+  // which the /live.js assembler serializes into these globals alongside the
+  // token/port/vocabulary. This file is served raw and injected as a classic
+  // script, so it cannot import that module; the private impeccable-site repo
+  // imports it directly to check its Live UI lab holds a snapshot for every
+  // surface, which only works while the list has exactly one definition.
+  // Add a surface in ui-surfaces.mjs, not here.
+  const LIVE_CHROME_MOUNT_CONTRACT = Array.isArray(window.__IMPECCABLE_LIVE_MOUNT_CONTRACT__)
+    ? window.__IMPECCABLE_LIVE_MOUNT_CONTRACT__
+    : ['root', 'transport', 'state', 'actions'];
+  const LIVE_UI_SURFACES = Array.isArray(window.__IMPECCABLE_LIVE_UI_SURFACES__)
+    ? window.__IMPECCABLE_LIVE_UI_SURFACES__
+    : [];
   const LIVE_UI_COMPONENT_IDS = [...new Set(LIVE_UI_SURFACES.flatMap((surface) => surface.ids))];
 
   //

.claude/skills/impeccable/scripts/live-copy-edit-agent.mjs

@@ -14,10 +14,12 @@ import path from 'node:path';
 import { createRequire } from 'node:module';
 
 const DEFAULT_TIMEOUT_MS = 60_000;
+const BATCH_OP_TEXT_LIMIT = 240;
 const require = createRequire(import.meta.url);
 
 export function buildCopyEditBatchPrompt(batch, { cwd = process.cwd() } = {}) {
-  const repairLines = batch?.repair ? [
+  const compactBatch = compactBatchForPrompt(batch);
+  const repairLines = compactBatch.repair ? [
     '',
     'Repair mode:',
     '- The previous Apply attempt changed source, but validation failed.',
@@ -28,7 +30,7 @@ export function buildCopyEditBatchPrompt(batch, { cwd = process.cwd() } = {}) {
     '- If failures or candidates show edited text is also a lookup key, update coupled count, animation, icon, image, asset, style, or metadata keys in the current source, or fail that entry without partial edits.',
     '- Keep failed and notes as arrays.',
     '- Return the same canonical JSON shape after repair.',
-    JSON.stringify(batch.repair, null, 2),
+    JSON.stringify(compactBatch.repair, null, 2),
   ] : [];
   return [
     'You are the Impeccable staged copy-edit batch applier.',
@@ -80,7 +82,7 @@ export function buildCopyEditBatchPrompt(batch, { cwd = process.cwd() } = {}) {
     ...repairLines,
     '',
     'Staged copy-edit batch:',
-    JSON.stringify(compactBatchForPrompt(batch), null, 2),
+    JSON.stringify(compactBatch, null, 2),
   ].join('\n');
 }
 
@@ -292,7 +294,7 @@ function readManualEditValidationScript(cwd) {
 function compactBatchForPrompt(batch) {
   return {
     pageUrl: batch?.pageUrl || null,
-    repair: batch?.repair || undefined,
+    repair: compactBatchRepair(batch?.repair),
     entries: (batch?.entries || []).map((entry) => ({
       id: entry.id,
       pageUrl: entry.pageUrl,
@@ -300,7 +302,71 @@ function compactBatchForPrompt(batch) {
       element: compactContextForBatch(entry.element),
       ops: (entry.ops || []).map(compactBatchOp),
     })),
-    candidates: batch?.candidates || [],
+    candidates: compactBatchCandidates(batch?.candidates),
+  };
+}
+
+function compactBatchRepair(repair) {
+  if (!repair || typeof repair !== 'object') return undefined;
+  return {
+    status: compactBatchString(repair.status),
+    attempt: normalizeOptionalBatchNumber(repair.attempt),
+    attempts: normalizeOptionalBatchNumber(repair.attempts),
+    maxAttempts: normalizeOptionalBatchNumber(repair.maxAttempts),
+    reason: compactBatchString(repair.reason),
+    transactionId: compactBatchString(repair.transactionId),
+    pageUrl: compactBatchString(repair.pageUrl),
+    failures: compactBatchDiagnostics(repair.failures),
+    files: compactBatchStringList(repair.files, 20),
+  };
+}
+
+function compactBatchDiagnostics(items, depth = 0) {
+  if (!Array.isArray(items)) return undefined;
+  return items.slice(0, 12).map((item) => ({
+    entryId: compactBatchString(item?.entryId || item?.id),
+    reason: compactBatchString(item?.reason || item?.kind),
+    detail: compactBatchString(item?.detail),
+    message: compactBatchString(item?.message),
+    file: compactBatchString(item?.file || item?.relativeFile),
+    line: normalizeOptionalBatchNumber(item?.line),
+    ref: compactBatchString(item?.ref),
+    marker: compactBatchString(item?.marker),
+    files: compactBatchStringList(item?.files, 8),
+    candidates: depth < 2 ? compactBatchSourceMatches(item?.candidates, 8) : undefined,
+    failures: depth < 2 ? compactBatchDiagnostics(item?.failures, depth + 1) : undefined,
+    checks: depth < 2 ? compactBatchDiagnostics(item?.checks, depth + 1) : undefined,
+  }));
+}
+
+function compactBatchCandidates(candidates) {
+  return (Array.isArray(candidates) ? candidates : [])
+    .slice(0, 24)
+    .map((candidate) => ({
+      entryId: compactBatchString(candidate?.entryId),
+      ref: compactBatchString(candidate?.ref),
+      sourceHint: compactBatchSourceMatch(candidate?.sourceHint),
+      textMatches: compactBatchSourceMatches(candidate?.textMatches, 8),
+      objectKeyMatches: compactBatchSourceMatches(candidate?.objectKeyMatches, 8),
+      contextTextMatches: compactBatchSourceMatches(candidate?.contextTextMatches, 8),
+      locatorMatches: compactBatchSourceMatches(candidate?.locatorMatches, 6),
+    }));
+}
+
+function compactBatchSourceMatches(matches, limit) {
+  if (!Array.isArray(matches)) return undefined;
+  return matches.slice(0, limit).map(compactBatchSourceMatch).filter(Boolean);
+}
+
+function compactBatchSourceMatch(match) {
+  if (!match || typeof match !== 'object') return null;
+  return {
+    file: compactBatchString(match.relativeFile || match.file),
+    line: normalizeBatchNumber(match.line),
+    column: normalizeBatchNumber(match.column),
+    kind: compactBatchString(match.kind),
+    reason: compactBatchString(match.reason || match.kind),
+    status: compactBatchString(match.status),
   };
 }
 
@@ -311,25 +377,77 @@ function compactBatchOp(op) {
     contextRef: op.contextRef,
     tag: op.tag,
     elementId: op.elementId,
-    classes: op.classes,
+    classes: compactBatchStringList(op.classes, 24),
     originalText: op.originalText,
     newText: op.newText,
     deleted: op.deleted === true || undefined,
-    sourceHint: op.sourceHint,
+    sourceHint: normalizeBatchSourceHint(op.sourceHint),
     leaf: compactContextForBatch(op.leaf),
-    nearbyEditableTexts: Array.isArray(op.nearbyEditableTexts) ? op.nearbyEditableTexts.slice(0, 8) : [],
+    nearbyEditableTexts: compactNearbyBatchTexts(op.nearbyEditableTexts),
     container: compactContextForBatch(op.container),
-    contextHints: Array.isArray(op.contextHints) ? op.contextHints.slice(0, 12) : [],
+    contextHints: compactBatchStringList(op.contextHints, 12),
   };
 }
 
+function normalizeBatchSourceHint(hint) {
+  if (!hint || typeof hint !== 'object') return null;
+  let line = normalizeBatchNumber(hint.line);
+  let column = normalizeBatchNumber(hint.column);
+  if ((line === null || column === null) && typeof hint.loc === 'string') {
+    const match = hint.loc.match(/^(\d+)(?::(\d+))?/);
+    if (match) {
+      line = Number(match[1]);
+      if (match[2]) column = Number(match[2]);
+    }
+  }
+  return {
+    file: compactBatchString(hint.file) || '',
+    loc: compactBatchString(hint.loc) || '',
+    line,
+    column,
+  };
+}
+
+function normalizeBatchNumber(value) {
+  if (value === null || value === undefined || value === '') return null;
+  const number = Number(value);
+  return Number.isFinite(number) ? number : null;
+}
+
+function normalizeOptionalBatchNumber(value) {
+  const number = normalizeBatchNumber(value);
+  return number === null ? undefined : number;
+}
+
+function compactNearbyBatchTexts(items) {
+  return (Array.isArray(items) ? items : [])
+    .slice(0, 8)
+    .map((item) => typeof item === 'string' ? { text: truncate(item, BATCH_OP_TEXT_LIMIT) } : {
+      ref: compactBatchString(item?.ref),
+      tag: compactBatchString(item?.tag),
+      classes: compactBatchStringList(item?.classes, 24),
+      text: compactBatchString(item?.text),
+    });
+}
+
+function compactBatchStringList(items, limit) {
+  return (Array.isArray(items) ? items : [])
+    .slice(0, limit)
+    .filter((item) => typeof item === 'string')
+    .map((item) => truncate(item, BATCH_OP_TEXT_LIMIT));
+}
+
+function compactBatchString(value) {
+  return typeof value === 'string' ? truncate(value, BATCH_OP_TEXT_LIMIT) : undefined;
+}
+
 function compactContextForBatch(value) {
   if (!value || typeof value !== 'object') return value || null;
   return {
-    ref: value.ref,
-    tagName: value.tagName,
-    id: value.id,
-    classes: value.classes,
+    ref: compactBatchString(value.ref),
+    tagName: compactBatchString(value.tagName),
+    id: compactBatchString(value.id),
+    classes: compactBatchStringList(value.classes, 24),
     textContent: truncate(value.textContent, 900),
     outerHTML: truncate(stripLiveRuntimeHtml(value.outerHTML), 1800),
   };
@@ -470,12 +588,11 @@ function runClaude(prompt, { cwd, env, resultPath, logPath, timeoutMs = DEFAULT_
   if (env.IMPECCABLE_LIVE_COPY_AGENT_MODEL) {
     args.push('--model', env.IMPECCABLE_LIVE_COPY_AGENT_MODEL);
   }
-  args.push(prompt);
   // Forward env as-is so CLAUDE_CODE_OAUTH_TOKEN and ANTHROPIC_API_KEY flow
   // through. On macOS, `claude /login` stores creds in the Keychain, which a
   // non-TTY subprocess cannot read; setting CLAUDE_CODE_OAUTH_TOKEN (via
   // `claude setup-token`) is the supported headless auth path.
-  return runAgentProcess('claude', args, '', { cwd, env, logPath, timeoutMs, mirrorOutputPath: resultPath });
+  return runAgentProcess('claude', args, prompt, { cwd, env, logPath, timeoutMs, mirrorOutputPath: resultPath });
 }
 
 function runAgentProcess(command, args, stdin, { cwd, env, logPath, timeoutMs, mirrorOutputPath }) {

.claude/skills/impeccable/scripts/live.mjs

@@ -17,7 +17,7 @@
  *   node live.mjs --help
  */
 
-import { execSync } from 'node:child_process';
+import { execFileSync } from 'node:child_process';
 import fs from 'node:fs';
 import path from 'node:path';
 import { fileURLToPath } from 'node:url';
@@ -316,11 +316,17 @@ function globToRegex(pattern) {
 
 function runScript(name, args, options = {}) {
   const scriptPath = path.join(__dirname, name);
-  const cmd = `node "${scriptPath}" ${args.map(a => `"${a}"`).join(' ')}`;
   try {
-    return execSync(cmd, { encoding: 'utf-8', cwd: options.cwd || process.cwd(), timeout: 15_000 });
+    // argv form, never a shell: string interpolation into double quotes would
+    // let a `"` or `$(...)` in any future caller's arg escape into the shell
+    // (issue #476).
+    return execFileSync(process.execPath, [scriptPath, ...args], {
+      encoding: 'utf-8',
+      cwd: options.cwd || process.cwd(),
+      timeout: 15_000,
+    });
   } catch (err) {
-    // execSync throws on non-zero exit; return stdout if any
+    // execFileSync throws on non-zero exit; return stdout if any
     return err.stdout || err.message || '';
   }
 }

.claude/skills/impeccable/scripts/live/browser-script-parts.mjs

@@ -1,6 +1,8 @@
 import fs from 'node:fs';
 import path from 'node:path';
 
+import { LIVE_CHROME_MOUNT_CONTRACT, LIVE_UI_SURFACES } from './ui-surfaces.mjs';
+
 export const LIVE_BROWSER_SCRIPT_PARTS = Object.freeze([
   Object.freeze({ name: 'session-state', file: 'live-browser-session.js' }),
   Object.freeze({ name: 'dom-helpers', file: 'live-browser-dom.js' }),
@@ -32,7 +34,20 @@ export function readLiveBrowserScriptParts(parts, readFile = (filePath) => fs.re
   }));
 }
 
-export function assembleLiveBrowserScript({ token, port, vocabulary, commandPrefix = '/', appRoot = null, parts }) {
+export function assembleLiveBrowserScript({
+  token,
+  port,
+  vocabulary,
+  commandPrefix = '/',
+  appRoot = null,
+  parts,
+  // Defaulted rather than threaded through live-server.mjs: the browser bundle
+  // must always carry the canonical inventory, and a default makes that true by
+  // construction instead of by every caller remembering to pass it. Overridable
+  // so tests can assemble with a stand-in.
+  uiSurfaces = LIVE_UI_SURFACES,
+  mountContract = LIVE_CHROME_MOUNT_CONTRACT,
+}) {
   const prelude =
     `window.__IMPECCABLE_TOKEN__ = '${token}';\n` +
     `window.__IMPECCABLE_PORT__ = ${port};\n` +
@@ -44,7 +59,14 @@ export function assembleLiveBrowserScript({ token, port, vocabulary, commandPref
     `window.__IMPECCABLE_COMMAND_PREFIX__ = ${JSON.stringify(commandPrefix)};\n` +
     // Canonical command vocabulary (values + labels + icons). live-browser.js
     // builds its action picker from this instead of an inline copy.
-    `window.__IMPECCABLE_VOCAB__ = ${JSON.stringify(vocabulary)};\n`;
+    `window.__IMPECCABLE_VOCAB__ = ${JSON.stringify(vocabulary)};\n` +
+    // Canonical Live chrome inventory from live/ui-surfaces.mjs. live-browser.js
+    // is a classic script and cannot import an ES module at runtime, so the list
+    // is serialized here and read off the global there. Node consumers (this
+    // repo's tests, the impeccable-site Live UI lab) import the module directly,
+    // which is what keeps the two from drifting.
+    `window.__IMPECCABLE_LIVE_UI_SURFACES__ = ${JSON.stringify(uiSurfaces)};\n` +
+    `window.__IMPECCABLE_LIVE_MOUNT_CONTRACT__ = ${JSON.stringify(mountContract)};\n`;
 
   const body = parts.map((part) => {
     const file = part.file || path.basename(part.path || '');

.claude/skills/impeccable/scripts/live/ui-surfaces.mjs

@@ -0,0 +1,75 @@
+/**
+ * Canonical inventory of the Live overlay's UI surfaces: one entry per piece of
+ * chrome Live mounts on the user's page, with the element ids that make it up.
+ *
+ * Single source of truth, consumed by:
+ *   - skill/scripts/live/browser-script-parts.mjs — serializes this into
+ *     window.__IMPECCABLE_LIVE_UI_SURFACES__ in the /live.js prelude.
+ *   - skill/scripts/live-browser.js — publishes it on
+ *     window.__IMPECCABLE_LIVE_CHROME_CORE__ for adapters and E2E probes. That
+ *     file is served raw and injected as a classic <script>, so it cannot
+ *     import this module at runtime; it reads the injected global instead, the
+ *     same path live/vocabulary.mjs already takes for the command palette.
+ *   - the private impeccable-site repo — site/components/LiveUiGallery.astro
+ *     and tests/live-ui-lab.test.mjs import LIVE_UI_SURFACES at build time and
+ *     fail the site build when the Live UI lab has no snapshot for a surface
+ *     defined here. That guard only guards if it reads this list rather than a
+ *     copy the site keeps, so this module must stay importable from Node.
+ *     Renaming a key or the module is a breaking change for that build; the
+ *     list was briefly inlined into live-browser.js and the site had to parse
+ *     it back out with a regex.
+ *
+ * Add a surface here and both the browser bundle and the site lab follow.
+ */
+
+/** Id prefix every Live chrome element carries. Mirrored by PREFIX in live-browser.js. */
+export const LIVE_UI_PREFIX = 'impeccable-live';
+
+const id = (suffix) => `${LIVE_UI_PREFIX}-${suffix}`;
+
+/**
+ * The mount contract every Live chrome adapter (DOM, Svelte, ...) satisfies.
+ * Published alongside the surfaces on __IMPECCABLE_LIVE_CHROME_CORE__.
+ */
+export const LIVE_CHROME_MOUNT_CONTRACT = Object.freeze(['root', 'transport', 'state', 'actions']);
+
+export const LIVE_UI_SURFACES = Object.freeze([
+  {
+    key: 'global-bottom-bar',
+    ids: [
+      id('global-bar'), id('global-bar-brand'), id('pick-toggle'), id('insert-toggle'),
+      id('detect-toggle'), id('detect-badge'), id('design-toggle'), id('page-chat'),
+      id('page-chat-input'), id('page-chat-voice'), id('page-chat-send'),
+    ],
+  },
+  { key: 'pending-copy-edit-dock', ids: [id('pending-dock')] },
+  {
+    key: 'element-selection-chrome',
+    ids: [
+      id('highlight'), id('tooltip'), id('bar'), id('selection-pill'), id('input'),
+      id('configure-voice'), id('configure-bar-tooltip'),
+    ],
+  },
+  { key: 'action-picker', ids: [id('picker')] },
+  { key: 'edit-chrome', ids: [id('edit-badge')] },
+  { key: 'generating-row', ids: [id('bar'), id('shader')] },
+  { key: 'variant-cycling-row', ids: [id('bar'), id('params-panel')] },
+  { key: 'variant-params-panel', ids: [id('params-panel')] },
+  { key: 'saving-confirmed-rows', ids: [id('bar')] },
+  {
+    key: 'insert-mode-chrome',
+    ids: [
+      id('insert-line'), id('insert-placeholder'), id('placeholder-resize'), id('insert-input'),
+      id('insert-voice'), id('insert-create'), id('insert-create-tooltip'),
+    ],
+  },
+  { key: 'annotation-chrome', ids: [id('annot'), id('annot-svg'), id('annot-pins'), id('annot-clear')] },
+  { key: 'design-system-panel', ids: [id('design-host')] },
+  { key: 'toasts-and-errors', ids: [id('toast'), id('mount-error')] },
+  { key: 'css-isolation-boundary', ids: [id('root')] },
+].map((surface) => Object.freeze({ ...surface, ids: Object.freeze(surface.ids) })));
+
+/** Every id any surface owns, de-duplicated, in surface order. */
+export const LIVE_UI_COMPONENT_IDS = Object.freeze([
+  ...new Set(LIVE_UI_SURFACES.flatMap((surface) => surface.ids)),
+]);

.claude/skills/impeccable/scripts/pin.mjs

@@ -93,15 +93,17 @@ function commandPrefixForSkillsDir(skillsDir) {
   return CODEX_HARNESSES.has(basename(dirname(skillsDir))) ? '$' : '/';
 }
 
-function generatePinnedSkill(command, metadata, commandPrefix) {
+function generatePinnedSkill(command, metadata, commandPrefix, isCodex) {
   const desc = metadata[command]?.description || `Shortcut for ${commandPrefix}impeccable ${command}.`;
   const hint = metadata[command]?.argumentHint || '[target]';
+  const providerFrontmatter = isCodex
+    ? `metadata:\n  argument-hint: "${hint}"`
+    : `argument-hint: "${hint}"\nuser-invocable: true`;
 
   return `---
 name: ${command}
 description: "${desc}"
-argument-hint: "${hint}"
-user-invocable: true
+${providerFrontmatter}
 ---
 
 ${PIN_MARKER}
@@ -128,7 +130,7 @@ function pin(command, projectRoot) {
 
   for (const skillsDir of harnessDirs) {
     const commandPrefix = commandPrefixForSkillsDir(skillsDir);
-    const content = generatePinnedSkill(command, metadata, commandPrefix);
+    const content = generatePinnedSkill(command, metadata, commandPrefix, commandPrefix === '$');
     // Check if skill already exists (and isn't a pin)
     const skillDir = join(skillsDir, command);
     if (existsSync(skillDir)) {

.claude/skills/impeccable/scripts/serve-question.mjs

@@ -29,25 +29,50 @@
  *       "materials": ["letterpress", "newsprint"],                // optional, rendered as tags
  *       "viewport": "one line: the first-viewport composition",   // optional
  *       "case": "one line: the fusion verdict, honest",           // optional
+ *       "verdict": "competitive", // optional routing tier: "wins" |
+ *                                // "competitive" | "declined". Declined cards
+ *                                // render demoted after the full cards:
+ *                                // narrow, quiet, catalog art as a labeled
+ *                                // thumb, "Adopt anyway" instead of "Build
+ *                                // this". Still choosable; never deleted.
+ *       "kept": "one line: what the direction kept from this declined world",
+ *       "raised": [ { "from": "challenger-x", "raise": "one line" } ],
+ *                                // assigned card only: donations taken from
+ *                                // declined challengers, rendered as named
+ *                                // raise lines under the identity row
  *       "risk": "one line: the honest risk",                      // optional
  *       "body": "fallback prose when the structured fields are absent",
- *       "sketch": ".impeccable/sketches/assigned.webp",  // optional; may not exist
- *                                // yet: the page shimmer-waits and polls the
- *                                // slot until the file lands, so serve first
- *                                // and generate after
+ *       "sketch": ".impeccable/mocks/decision/assigned.webp",  // optional; the card's
+ *                                // full-fidelity direction comp (the field
+ *                                // keeps the sketch era's wire name). May not
+ *                                // exist yet: the page shimmer-waits and
+ *                                // polls the slot until the file lands, so
+ *                                // serve first and generate after
  *       "hero": "https://... or /abs/path.webp",   // optional inspiration image;
  *                                // rides picture-in-picture when a sketch exists
  *       "board": "https://... or /abs/path.webp"   // optional secondary image
  *     }, ...
  *   ],
  *   "reroll": true,          // adds a re-roll action (returns {"optionId":"reroll"})
+ *                            // or { "registers": ["safer", "bolder"] } to add
+ *                            // the register steers beside it: the answer then
+ *                            // carries "register" and the agent re-runs
+ *                            // concept-seed with --register <value>
  *   "canon": true,           // adds the "Play it straight" standing exit;
  *                            // direction rounds only (returns {"optionId":"canon"})
  *   "canonCard": { ... },    // optional: the standing exit as a full card with the
  *                            // same anatomy (label, thesis, palette, sketch, ...);
  *                            // rendered last and visually subordinate. Without it,
  *                            // canon stays a quiet footer action.
- *   "steer": true            // adds a free-text steer field returned with any answer
+ *   "steer": true,           // adds a free-text steer field returned with any answer
+ *   "followup": true         // this round's pick is not terminal: the server
+ *                            // stays open awaiting --update with the next
+ *                            // round (detached mode only), the page shows a
+ *                            // loading hand instead of goodbye, and the
+ *                            // answer carries followup:true so --wait knows
+ *                            // to keep the table. Use it when a decision has
+ *                            // a known second half, e.g. direction first,
+ *                            // then the execution contract.
  * }
  *
  * Options render as large cards: the sketch leads when present, with the
@@ -79,6 +104,7 @@ import fs from 'node:fs';
 import path from 'node:path';
 import { spawn } from 'node:child_process';
 import { fileURLToPath } from 'node:url';
+import { openSystemBrowser } from './lib/open-system-browser.mjs';
 
 function arg(name, fallback = null) {
   const i = process.argv.indexOf(`--${name}`);
@@ -123,11 +149,17 @@ function printAnswer(raw) {
       console.log("CHOSEN CARD: open the chosen world's board and hero images now, before any code. When your harness only reads files, or runs sandboxed, download them INTO the workspace and open the relative path; a sandboxed viewer rejects absolute paths outside it. They set the craft bar the build must reach.");
     }
     if (a.sketch) {
-      console.log('CHOSEN SKETCH: the decision sketch at that path may seed one comp probe; the comp round still renders its full set, because a sketch chose the direction, not the composition.');
+      console.log('CHOSEN COMP: the decision comp at that path is compositional option one. On a comp-led build the comp round adds two variations beside it; on a code-led build it returns at the finish review as the critique reference. Never regenerate it from scratch.');
     }
     if (a.optionId === 'canon') {
       console.log('CANON CHOSEN: the user picked the category standard on purpose. Ask once for two or three products this should sit alongside; their craft level becomes the quality bar. Execute the canon at full commitment, conventions embraced without irony or smuggled quirk.');
     }
+    if (a.optionId === 'reroll' && a.register) {
+      console.log(`REGISTER: the user steered the next hand to the ${a.register} register. Re-run concept-seed with the same key, the next --reroll round, and --register ${a.register}, then follow what it prints; the register is the user's steering, never yours to pre-select.`);
+    }
+    if (a.followup && a.optionId !== 'reroll') {
+      console.log('FOLLOWUP OPEN: the table stays open and the page is showing a loading hand. Deliver the next round now with --update --key <key> --payload <file>, then collect it with --wait; never leave the page waiting on a round you have not sent.');
+    }
   } catch { /* raw answer */ }
 }
 
@@ -143,15 +175,17 @@ if (hasFlag('schema')) {
     title: 'Choose the visual world',
     question: 'The roll assigned Fillmore Handbill. Keep it, take an alternate, or re-roll.',
     options: [
-      { id: 'assigned', label: 'Fillmore Handbill', kicker: 'THE ROLL', lineage: '1966-71 Fillmore psychedelic handbills', thesis: 'The gig poster that treats every release like a one-night stand.', palette: ['#e8452c', '#f5d64c', '#1b2a52', '#f3ead8'], materials: ['letterpress', 'split-fountain ink'], viewport: 'A full-bleed dated bill with the product name in warped display type.', risk: 'Reads nostalgic when the type is set timidly.', sketch: '.impeccable/sketches/assigned.webp', hero: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill-hero.webp', board: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill.webp' },
-      { id: 'challenger-teletext', label: 'Teletext Service', lineage: 'broadcast teletext magazines', thesis: 'The catalog as a broadcast index: pages, not sections.', case: 'Fuses cleanly: releases map to numbered pages.', sketch: '.impeccable/sketches/challenger-teletext.webp', hero: 'https://impeccable.style/worlds/cards/broadcast-programming-teletext-service-hero.webp' },
+      { id: 'assigned', label: 'Fillmore Handbill', kicker: 'THE ROLL', lineage: '1966-71 Fillmore psychedelic handbills', thesis: 'The gig poster that treats every release like a one-night stand.', palette: ['#e8452c', '#f5d64c', '#1b2a52', '#f3ead8'], materials: ['letterpress', 'split-fountain ink'], viewport: 'A full-bleed dated bill with the product name in warped display type.', risk: 'Reads nostalgic when the type is set timidly.', raised: [{ from: 'challenger-microfiche', raise: 'The bill now owns its whole viewport as one continuous printed sheet.' }], sketch: '.impeccable/mocks/decision/assigned.webp', hero: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill-hero.webp', board: 'https://impeccable.style/worlds/cards/posters-covers-sleeves-fillmore-handbill.webp' },
+      { id: 'model-pick', label: 'The Broadside Ballad', kicker: 'MY PICK', lineage: 'street-sold ballad sheets', thesis: 'Every release printed as the day’s ballad sheet.', palette: ['#1f1c18', '#efe5d0', '#a33327'], materials: ['woodcut', 'rag paper'], viewport: 'One tall sheet, the newest release as today’s ballad.', risk: 'Also the direction most runs in this category land on.', sketch: '.impeccable/mocks/decision/model-pick.webp' },
+      { id: 'challenger-teletext', label: 'Teletext Service', verdict: 'competitive', lineage: 'broadcast teletext magazines', thesis: 'The catalog as a broadcast index: pages, not sections.', palette: ['#0000c0', '#ffff00', '#00c000', '#ffffff'], materials: ['block mosaic', 'phosphor glow'], viewport: 'P100 index page, releases as numbered rows.', case: 'Fuses cleanly: releases map to numbered pages; loses narrowly on clarity.', risk: 'Reads retro-novelty when the grid is not strict.', sketch: '.impeccable/mocks/decision/challenger-teletext.webp', hero: 'https://impeccable.style/worlds/cards/broadcast-programming-teletext-service-hero.webp' },
+      { id: 'challenger-microfiche', label: 'Microfiche Reader', verdict: 'declined', lineage: 'library microfiche stations', palette: ['#101418', '#9fb4c0'], materials: ['film grain', 'backlit glass'], case: 'Fuses poorly: listeners do not identify with archival retrieval.', kept: 'Total environmental commitment.', hero: 'https://impeccable.style/worlds/cards/archives-microfiche-reader-hero.webp' },
     ],
-    reroll: true,
+    reroll: { registers: ['safer', 'bolder'] },
     canon: true,
-    canonCard: { label: 'The category standard', thesis: 'What this category ships, executed impeccably.', viewport: 'The arrangement a visitor expects, at full craft.', sketch: '.impeccable/sketches/canon.webp' },
+    canonCard: { label: 'The category standard', thesis: 'What this category ships, executed impeccably.', palette: ['#ffffff', '#111827', '#2563eb'], materials: ['clean grid', 'product photography'], viewport: 'The arrangement a visitor expects, at full craft.', risk: 'Indistinguishable from the competition by design.', sketch: '.impeccable/mocks/decision/canon.webp' },
     steer: true,
   }, null, 2));
-  console.log('\nOption ids return verbatim in ANSWER; "reroll" and "canon" are reserved. hero/board/sketch accept URLs or local paths; sketch slots may point at files that do not exist yet (serve first, generate after; the page polls until they land, so never block serving on generation). hero on a challenger is the inspiration it draws from and renders picture-in-picture beside the sketch, never as the promise of the build. canonCard renders the standing exit as a subordinate card with the same anatomy; without it, canon stays a quiet footer action. Include canon only for visual-direction rounds; never present it as your own recommendation. Keep thesis and each fact to one short sentence: the card front shows thesis, identity, and a two-line risk, while first viewport and the case read on the card back behind the Details chip, so long facts cost the reader a flip, not the page its scanability. A card with no imagery at all has no back; its full read renders on the front, so a text-only round loses nothing. Sketch aspect follows the surface: portrait at device viewport for native or mobile-first surfaces, landscape otherwise; the page adapts its cards to either.');
+  console.log('\nOption ids return verbatim in ANSWER; "reroll" and "canon" are reserved. hero/board/sketch accept URLs or local paths; sketch slots may point at files that do not exist yet (serve first, generate after; the page polls until they land, so never block serving on generation). hero on a challenger is the inspiration it draws from and renders picture-in-picture beside the sketch, never as the promise of the build. verdict routes rendering: "wins" and "competitive" challengers keep full cards, "declined" ones render demoted after them (narrow, quiet, art as a labeled thumb, "Adopt anyway"), with their kept line on the front; the page reorders declined cards to the end on its own. raised on the assigned card renders each donation as a named raise line. Salience parity: when the assigned card declares no sketch (no image generation this round), catalog art on every card demotes to a labeled thumb, so what looks important is the verdict’s call, never rendering luck. canonCard renders the standing exit as a subordinate card with the same anatomy; without it, canon stays a quiet footer action. Include canon only for visual-direction rounds; never present it as your own recommendation. The pick card is a kicker convention, not a field: kicker "MY PICK" on your top-ranked grounded candidate, one at most, never in the lead slot. Every card gets the full anatomy, challengers, canon, and declined included: thesis, palette, materials, viewport, risk; the seed already hands you each challenger’s system rules, so a card with no palette chips is an authoring gap, not a data gap. Keep thesis and each fact to one short sentence: the card front shows thesis, identity, and a two-line risk, while first viewport and the case read on the card back behind the Details chip, so long facts cost the reader a flip, not the page its scanability. A card with no imagery at all has no back; its full read renders on the front, so a text-only round loses nothing. The sketch slot carries the card’s full-fidelity direction comp (the field keeps its wire name for compatibility). Comp aspect follows the surface: portrait at device viewport for native or mobile-first surfaces, landscape otherwise; the page adapts its cards to either. reroll accepts true or { "registers": ["safer", "bolder"] }: the register buttons steer the next hand along the familiar-to-bold axis, the answer carries "register", and you re-run concept-seed with --register <value> for the next round; offer the registers on direction rounds, and never pre-select one. followup: true keeps the table open after a pick for a second round via --update (direction first, then the execution contract); send the next payload immediately, the page is waiting on it.');
   process.exit(0);
 }
 
@@ -195,12 +229,16 @@ if (hasFlag('wait')) {
   if (!answered()) { console.log(`WAITING: no answer yet after ${pollSec}s; run --wait --key ${key} again`); process.exit(3); }
   const collected = fs.readFileSync(answerFile(key), 'utf8').trim();
   printAnswer(collected);
-  // A re-roll keeps the table open: the server stays alive awaiting --update,
-  // so only the answer file is consumed. Terminal choices clean up fully.
-  let isRerollAnswer = false;
-  try { isRerollAnswer = JSON.parse(collected).optionId === 'reroll'; } catch { /* treat as terminal */ }
+  // A re-roll or a followup-round pick keeps the table open: the server stays
+  // alive awaiting --update, so only the answer file is consumed. Terminal
+  // choices clean up fully.
+  let keepsTableOpen = false;
+  try {
+    const parsedAnswer = JSON.parse(collected);
+    keepsTableOpen = parsedAnswer.optionId === 'reroll' || parsedAnswer.followup === true;
+  } catch { /* treat as terminal */ }
   try { fs.rmSync(answerFile(key)); } catch { /* already gone */ }
-  if (!isRerollAnswer) { try { fs.rmSync(stateFile(key)); } catch { /* already gone */ } }
+  if (!keepsTableOpen) { try { fs.rmSync(stateFile(key)); } catch { /* already gone */ } }
   process.exit(0);
 }
 
@@ -301,11 +339,19 @@ function loadRound(json) {
     sketchSrc: sketchSrc(option.sketch),
   });
   options = parsed.options.map(decorate);
+  // The verdict routes rendering: full cards first, then the canon, then the
+  // declined cards dead last in their own payload order. The reorder happens
+  // here so a payload that interleaves them still renders the weighing's
+  // shape, and the deck reads as a gradient of standing: contenders, the
+  // familiar door, then the demoted row.
+  const declined = options.filter((o) => o.verdict === 'declined');
+  options = options.filter((o) => o.verdict !== 'declined');
   // The standing exit as a full card: same anatomy, reserved id, rendered
   // subordinate by the page. Without it, canon stays the quiet footer action.
   if (parsed.canonCard && typeof parsed.canonCard === 'object') {
     options = [...options, { ...decorate(parsed.canonCard), id: 'canon', isCanon: true }];
   }
+  options = [...options, ...declined];
 }
 try { loadRound(raw); } catch (error) { console.error(`serve-question: ${error.message}`); process.exit(1); }
 const detachedKey = hasFlag('detached-serve') ? arg('key') : null;
@@ -321,7 +367,19 @@ function page() {
   // and material tags give a text-only direction an immediate identity that
   // no generation luck can distort.
   const fact = (label, value, cls = '') => value ? `<p class="fact${cls ? ` ${cls}` : ''}"><span class="fact-label">${label}</span>${esc(value)}</p>` : '';
-  const hasMedia = (option) => Boolean(option.sketchSrc || option.heroSrc || option.boardSrc);
+  const demoted = (option) => option.verdict === 'declined';
+  // Salience parity: a card's imagery weight is capped by the assigned card's.
+  // When the lead card has no media at all (no image generation this round,
+  // and no catalog art of its own), full-bleed catalog art beside a text-only
+  // assigned card would let rendering luck outvote the weighing: users click
+  // the colorful thing. Declined cards are thumb-only regardless; the verdict
+  // demoted them, and a full-bleed hero would promote them right back.
+  const identityRound = !(options[0] && (options[0].sketchSrc || options[0].heroSrc || options[0].boardSrc));
+  // A declined card never renders a full media face, sketch included: even a
+  // declared sketch would buy back the salience the verdict took away.
+  const faceSketch = (option) => demoted(option) ? null : option.sketchSrc;
+  const thumbOnly = (option) => !faceSketch(option) && Boolean(option.heroSrc || option.boardSrc) && (demoted(option) || identityRound);
+  const hasMedia = (option) => Boolean(faceSketch(option) || ((option.heroSrc || option.boardSrc) && !thumbOnly(option)));
   // The back exists to keep long facts off a card whose front is an image;
   // a card with no art has no flip chip to reach it, so it gets no back and
   // the full read lives on the front instead.
@@ -337,6 +395,30 @@ function page() {
       idBits.push(option.materials.slice(0, 4).map((m) => `<span class="tag">${esc(m)}</span>`).join(''));
     }
     if (idBits.length) rows.push(`<div class="identity">${idBits.join('')}</div>`);
+    // Donations from declined challengers render as named raise lines: the
+    // assigned card arrives already raised by the hand it beat, and the raise
+    // is readable, because a raise nobody can read did not happen. One raise
+    // renders inline; several become a compact cycler (click advances), so a
+    // generous hand cannot blow the card out of proportion.
+    if (Array.isArray(option.raised) && option.raised.length) {
+      const nameOf = (id) => options.find((o) => o.id === id)?.label || String(id ?? '');
+      const raiseLines = option.raised.slice(0, 6).map((r) => `<p class="raise"><span class="fact-label">Raised by ${esc(nameOf(r.from))}</span>${esc(r.raise || r.kept || '')}</p>`);
+      if (raiseLines.length > 1) {
+        rows.push(`<div class="raises raises-cycle" role="button" tabindex="0" title="Click or press Enter to see the next raise" aria-label="Raised by the hand; activate to see the next raise">
+              <div class="raises-head"><span class="fact-label">Raised by the hand</span><span class="raises-count" data-raises-count>1/${raiseLines.length}</span></div>
+              ${raiseLines.join('')}
+              <span class="sr-live" aria-live="polite"></span>
+            </div>`);
+      } else {
+        rows.push(`<div class="raises">${raiseLines[0]}</div>`);
+      }
+    }
+    // Demoted art stays reachable as a labeled thumb: the catalog world
+    // explains where the direction comes from without buying it back the
+    // salience the verdict took away.
+    if (thumbOnly(option)) {
+      rows.push(`<figure class="inspo" title="Inspiration: the world this direction draws from. Your page will not look like this image."><img src="${esc(option.heroSrc || option.boardSrc)}" alt=""><figcaption>inspired by</figcaption></figure>`);
+    }
     // The front carries only what the choice needs: thesis, identity, and the
     // honest risk clamped to two lines. First viewport and the case read on
     // the card's back; once the sketch lands, the first viewport is a picture.
@@ -347,6 +429,7 @@ function page() {
     } else {
       rows.push(fact('First viewport', option.viewport));
       rows.push(fact('The case', option.case));
+      rows.push(fact('Kept', option.kept));
       rows.push(fact('Risk', option.risk));
     }
     if (!option.thesis && option.body) rows.push(`<p class="detail">${esc(option.body)}</p>`);
@@ -356,6 +439,7 @@ function page() {
   const backFacts = (option) => [
     fact('First viewport', option.viewport),
     fact('The case', option.case),
+    fact('Kept', option.kept),
     fact('Risk', option.risk),
     option.body && option.thesis ? `<p class="detail more">${esc(option.body)}</p>` : '',
   ].filter(Boolean).join('\n            ');
@@ -365,9 +449,12 @@ function page() {
               <figcaption>inspiration</figcaption>
             </figure>` : '';
     const details = hasBack(option) ? flipChip('Details') : '';
-    if (option.sketchSrc) {
+    // Thumb-only art renders inside the body via anatomy(), never as a face,
+    // and a declined card's sketch slot is ignored outright.
+    if (thumbOnly(option)) return '';
+    if (faceSketch(option)) {
       return `<div class="media sketching" data-sketch="${esc(option.sketchSrc)}">
-            <div class="shimmer"><span class="sketch-note">sketching&hellip;</span></div>
+            <div class="shimmer"><span class="sketch-note">rendering&hellip;</span></div>
             <img class="sketch" alt="" hidden>
             ${inspiration}
             <div class="chips">${expandChip}${details}</div>
@@ -384,17 +471,18 @@ function page() {
     }
     return '';
   };
+  const chooseLabel = (option) => option.isCanon ? 'Play it straight' : demoted(option) ? 'Adopt anyway' : 'Build this';
   const cards = options.map((option, index) => `
-    <article class="card${option.isCanon ? ' canon' : ''}" style="--fan:${index === 0 ? '0deg' : (index % 2 ? '1.4deg' : '-1.2deg')};--deal:${index * 90}ms" data-id="${esc(option.id)}">
+    <article class="card${option.isCanon ? ' canon' : ''}${demoted(option) ? ' declined' : ''}" style="--fan:${index === 0 ? '0deg' : (index % 2 ? '1.4deg' : '-1.2deg')};--deal:${index * 90}ms" data-id="${esc(option.id)}">
       <div class="card-inner">
         <div class="face front${index === 0 ? ' lead' : ''}${media(option) ? '' : ' text-only'}">
-          ${option.kicker ? `<span class="kicker">${esc(option.kicker)}</span>` : option.isCanon ? '<span class="kicker standing">The standing door</span>' : ''}
+          ${option.kicker ? `<span class="kicker">${esc(option.kicker)}</span>` : demoted(option) ? '<span class="kicker declined-k">Declined</span>' : option.isCanon ? '<span class="kicker standing">The standing door</span>' : ''}
           ${media(option)}
           <div class="body">
             ${option.lineage ? `<p class="tier">${esc(option.lineage)}</p>` : ''}
             <h2>${esc(option.label)}</h2>
             ${anatomy(option)}
-            <button class="choose" data-id="${esc(option.id)}">${option.isCanon ? 'Play it straight' : 'Build this'}</button>
+            <button class="choose" data-id="${esc(option.id)}">${chooseLabel(option)}</button>
           </div>
         </div>
         ${hasBack(option) ? `<div class="face back${index === 0 ? ' lead' : ''}">
@@ -405,7 +493,7 @@ function page() {
           <div class="body back-body">
             ${option.boardSrc ? `<p class="tier">The full read &middot; ${esc(option.label)}</p>` : ''}
             ${backFacts(option)}
-            <button class="choose" data-id="${esc(option.id)}">${option.isCanon ? 'Play it straight' : 'Build this'}</button>
+            <button class="choose" data-id="${esc(option.id)}">${chooseLabel(option)}</button>
           </div>
         </div>` : ''}
       </div>
@@ -496,6 +584,10 @@ function page() {
     .nav.next { right: auto; left: 50%; top: auto; bottom: 6px; transform: translate(-50%, 0); }
     .fade-prev { top: 0; left: 0; right: 0; bottom: auto; width: auto; height: 72px; background: linear-gradient(180deg, var(--ks-lacquer), transparent); }
     .fade-next { top: auto; left: 0; right: 0; bottom: 0; width: auto; height: 72px; background: linear-gradient(0deg, var(--ks-lacquer), transparent); }
+    /* In the vertical deck the cross axis is horizontal: flex-start would
+       shrink a declined card to content WIDTH, not height, so it stretches
+       like every other card and its height is already its own. */
+    .grid > .card.declined { align-self: stretch; }
   }
   .card { position: relative; perspective: 1400px; transform: rotate(var(--fan, 0deg)); transition: transform .25s cubic-bezier(.16, 1, .3, 1); }
   .card:hover { transform: rotate(0deg) translateY(-4px); }
@@ -584,6 +676,42 @@ function page() {
   /* The generic .media img display:block would defeat [hidden] and float an
      empty block over the shimmer; an unloaded sketch must truly not render. */
   .media img[hidden] { display: none; }
+  /* Declined challengers: the weighing demoted them, so the card is narrower
+     and quieter, its catalog art rides as a labeled thumb in the body, and
+     the action reads "Adopt anyway". Adoptable, never deleted: the demoted
+     row is the hand's proof of judgment. */
+  /* Narrow AND short: without align-self the stretch default drags a thin
+     declined card to the tallest contender's height, a strange stilt of a
+     card beside the full hand. */
+  .grid > .card.declined { flex: 0 0 clamp(15rem, 21vw, 21rem); align-self: flex-start; }
+  .card.declined .face { background: var(--ks-graphite); }
+  .card.declined:hover .face { border-color: var(--ks-text-faint); }
+  .card.declined h2 { font-size: 1rem; color: var(--ks-text); }
+  .kicker.declined-k { background: transparent; border: 1px solid var(--ks-rule); color: var(--ks-text-faint); }
+  .card.declined button.choose { background: transparent; color: var(--ks-text-muted); border: 1px solid var(--ks-rule); font-size: .85rem; padding: 8px 22px; }
+  .card.declined button.choose:hover { background: var(--ks-graphite-2); border-color: var(--ks-text-muted); }
+  /* Thumb-scale inspiration: present, labeled, zoomable, and incapable of
+     outshouting a text-only assigned card. */
+  .inspo { position: relative; flex: none; margin: 2px 0; width: 104px; height: 64px; border: 1px solid var(--ks-rule); border-radius: 6px; overflow: hidden; cursor: zoom-in; background: var(--ks-lacquer); }
+  .inspo img { display: block; width: 100%; height: 100%; object-fit: cover; }
+  .inspo figcaption { position: absolute; left: 0; right: 0; bottom: 0; font-family: var(--ks-mono); font-size: .48rem; letter-spacing: .16em; text-transform: uppercase; color: var(--ks-text); text-align: center; padding: 2px 0 3px; background: oklch(7% 0.006 95 / 0.72); }
+  /* Raises: the donations the assigned direction took from the hand it beat,
+     each named for its donor. Patina, not kinpaku: a raise is provenance. */
+  .raises { display: flex; flex-direction: column; gap: 4px; margin: 2px 0; }
+  .raise { font-size: .78rem; color: var(--ks-text-muted); line-height: 1.45; border-left: 2px solid var(--ks-patina); padding-left: 8px; }
+  .raise .fact-label { color: var(--ks-patina); }
+  /* Several raises cycle instead of stacking: one visible at a time, a
+     counter for the rest, the whole block advances on click. */
+  .raises-cycle { cursor: pointer; border-radius: 6px; }
+  .raises-cycle .raise { display: none; border-left: none; padding-left: 0; }
+  .raises-cycle .raise.active { display: block; }
+  .raises-cycle { border-left: 2px solid var(--ks-patina); padding-left: 8px; }
+  .raises-head { display: flex; align-items: baseline; justify-content: space-between; gap: 8px; }
+  .raises-head .fact-label { color: var(--ks-patina); }
+  .raises-count { font-family: var(--ks-mono); font-size: .58rem; letter-spacing: .14em; color: var(--ks-text-faint); }
+  .raises-count::after { content: " \\203A"; }
+  .raises-cycle:hover .raises-count { color: var(--ks-patina); }
+  .sr-live { position: absolute; width: 1px; height: 1px; padding: 0; margin: -1px; overflow: hidden; clip: rect(0 0 0 0); white-space: nowrap; border: 0; }
   /* The standing exit as a card: present with full anatomy, never dressed as a
      contender. Graphite instead of kinpaku, and it never takes the lead ring. */
   .card.canon .face { border-color: var(--ks-rule); background: var(--ks-graphite); }
@@ -596,9 +724,14 @@ function page() {
   footer { width: 100%; max-width: 90rem; margin: 1.6rem auto 0; display: flex; gap: 1rem; align-items: center; flex-wrap: wrap; }
   #steer { flex: 1; min-width: 16rem; background: var(--ks-lacquer-raised); color: var(--ks-text); border: 1px solid var(--ks-rule); border-radius: 7px; padding: .6rem .85rem; font: inherit; }
   #steer:focus { outline: none; border-color: var(--ks-patina); }
-  #reroll { display: inline-flex; align-items: center; align-self: stretch; gap: 8px; padding: 0 16px; font-family: var(--ks-mono); font-size: .72rem; letter-spacing: .08em; text-transform: uppercase; color: var(--ks-kinpaku); background: transparent; border: 1px solid var(--ks-rule); border-radius: 6px; cursor: pointer; transition: border-color .2s ease, color .2s ease; }
-  #reroll:hover { color: var(--ks-kinpaku-pale); border-color: var(--ks-kinpaku-deep); }
-  #reroll svg { width: 15px; height: 15px; }
+  .reroll-btn { display: inline-flex; align-items: center; align-self: stretch; gap: 8px; padding: 0 16px; font-family: var(--ks-mono); font-size: .72rem; letter-spacing: .08em; text-transform: uppercase; color: var(--ks-kinpaku); background: transparent; border: 1px solid var(--ks-rule); border-radius: 6px; cursor: pointer; transition: border-color .2s ease, color .2s ease; }
+  .reroll-btn:hover { color: var(--ks-kinpaku-pale); border-color: var(--ks-kinpaku-deep); }
+  .reroll-btn svg { width: 15px; height: 15px; }
+  .reroll-btn[disabled] { opacity: .4; cursor: default; }
+  /* The register steers read quieter than the plain roll: they are exits from
+     the current register, not the round's main verbs. */
+  #reroll-safer, #reroll-bolder { color: var(--ks-text-muted); min-height: 38px; }
+  #reroll-safer:hover, #reroll-bolder:hover { color: var(--ks-text); border-color: var(--ks-text-faint); }
   /* The quiet exit: always available, never argued with, visually subordinate
      to the dealt cards and the re-roll so it reads as the user's own door,
      not a recommendation. */
@@ -643,16 +776,33 @@ function page() {
 </main>
 <footer>
   ${payload.steer ? '<input id="steer" placeholder="Optional steer: what should be different or kept?">' : ''}
-  ${payload.reroll ? '<button id="reroll"><svg viewBox="0 0 24 24" aria-hidden="true"><rect x="3" y="3" width="18" height="18" rx="4" fill="none" stroke="currentColor" stroke-width="1.6"/><circle cx="8.4" cy="8.4" r="1.5" fill="currentColor"/><circle cx="15.6" cy="8.4" r="1.5" fill="currentColor"/><circle cx="8.4" cy="15.6" r="1.5" fill="currentColor"/><circle cx="15.6" cy="15.6" r="1.5" fill="currentColor"/><circle cx="12" cy="12" r="1.5" fill="currentColor"/></svg><span>Re-roll</span></button>' : ''}
+  ${(() => {
+    if (!payload.reroll) return '';
+    const die = '<svg viewBox="0 0 24 24" aria-hidden="true"><rect x="3" y="3" width="18" height="18" rx="4" fill="none" stroke="currentColor" stroke-width="1.6"/><circle cx="8.4" cy="8.4" r="1.5" fill="currentColor"/><circle cx="15.6" cy="8.4" r="1.5" fill="currentColor"/><circle cx="8.4" cy="15.6" r="1.5" fill="currentColor"/><circle cx="15.6" cy="15.6" r="1.5" fill="currentColor"/><circle cx="12" cy="12" r="1.5" fill="currentColor"/></svg>';
+    const registers = Array.isArray(payload.reroll.registers) ? payload.reroll.registers.filter((r) => r === 'safer' || r === 'bolder') : [];
+    // The registers are the user's steering wheel on the familiar-to-bold
+    // axis; the plain re-roll sits between them so the spatial order matches
+    // the axis it names.
+    const safer = registers.includes('safer') ? '<button class="reroll-btn" id="reroll-safer" title="Deal the familiar register: conventional grounded directions plus the category standard against named competitors"><span>&larr; Safer hand</span></button>' : '';
+    const bolder = registers.includes('bolder') ? '<button class="reroll-btn" id="reroll-bolder" title="Deal foreign forms only, at full commitment"><span>Bolder hand &rarr;</span></button>' : '';
+    return `${safer}<button class="reroll-btn" id="reroll">${die}<span>Re-roll</span></button>${bolder}`;
+  })()}
   ${payload.canon && !payload.canonCard ? '<button id="canon" title="Skip the roll: build the page this category ships, executed impeccably">Play it straight</button>' : ''}
 </footer>
 <script>
   const steer = () => document.getElementById('steer')?.value || '';
+  // A followup round's pick keeps the tab: the next round arrives via
+  // --update, so the page shows the loading hand instead of goodbye. Detached
+  // mode only, and the page must agree with the server: a blocking server
+  // exits on any pick and has no update channel, so a followup payload there
+  // still gets the goodbye screen, never a loading hand nothing will resolve.
+  const FOLLOWUP = ${payload.followup === true && Boolean(detachedKey) ? 'true' : 'false'};
   const beat = () => { try { navigator.sendBeacon('/heartbeat'); } catch { fetch('/heartbeat', { method: 'POST' }); } };
   beat();
   setInterval(beat, 5000);
   async function answer(optionId) {
     await fetch('/answer', { method: 'POST', headers: { 'content-type': 'application/json' }, body: JSON.stringify({ optionId, steer: steer() }) });
+    if (FOLLOWUP) { await awaitNextRound(); return; }
     document.body.innerHTML = '<div class="done"><svg viewBox="0 0 24 24" width="38" height="38" fill="oklch(84% 0.19 80.46)" aria-hidden="true"><path d="M5 2.5 L13.5 2.5 L5.5 21.5 L5 21.5 Q2.5 21.5 2.5 19 L2.5 5 Q2.5 2.5 5 2.5 Z"/><path d="M16.5 2.5 L19 2.5 Q21.5 2.5 21.5 5 L21.5 19 Q21.5 21.5 19 21.5 L8.5 21.5 Z"/></svg>Choice recorded. The agent is resuming; you can close this tab.</div>';
   }
   document.querySelectorAll('button.choose').forEach(b => b.addEventListener('click', () => answer(b.dataset.id)));
@@ -661,6 +811,25 @@ function page() {
     b.closest('.card').classList.toggle('flipped');
   }));
 
+  // Raise cycler: click (or Enter) advances to the next donation.
+  document.querySelectorAll('.raises-cycle').forEach(cycle => {
+    const raises = [...cycle.querySelectorAll('.raise')];
+    const count = cycle.querySelector('[data-raises-count]');
+    let at = 0;
+    const live = cycle.querySelector('.sr-live');
+    const show = (announce) => {
+      raises.forEach((raise, i) => raise.classList.toggle('active', i === at));
+      if (count) count.textContent = (at + 1) + '/' + raises.length;
+      // Screen readers hear the raise they just advanced to; the initial
+      // render stays quiet so page load does not narrate every card.
+      if (announce && live) live.textContent = 'Raise ' + (at + 1) + ' of ' + raises.length + ': ' + (raises[at]?.textContent || '');
+    };
+    show(false);
+    const advance = (e) => { e.stopPropagation(); at = (at + 1) % raises.length; show(true); };
+    cycle.addEventListener('click', advance);
+    cycle.addEventListener('keydown', (e) => { if (e.key === 'Enter' || e.key === ' ') { e.preventDefault(); advance(e); } });
+  });
+
   // Deal from the stack: cards begin piled at the grid's center, blurred,
   // then travel to their seats with a stagger.
   const cards = [...document.querySelectorAll('.card')];
@@ -708,7 +877,7 @@ function page() {
     const note = m.querySelector('.sketch-note');
     const started = Date.now();
     // A live elapsed count is the difference between "working" and "frozen".
-    const tick = setInterval(() => { if (note) note.textContent = 'sketching · ' + Math.round((Date.now() - started) / 1000) + 's'; }, 1000);
+    const tick = setInterval(() => { if (note) note.textContent = 'rendering · ' + Math.round((Date.now() - started) / 1000) + 's'; }, 1000);
     const settle = () => { clearInterval(tick); m.classList.remove('sketching', 'stand-in'); m.querySelector('.shimmer')?.remove(); m.querySelector('.stand-in-label')?.remove(); };
     const standIn = () => {
       const pip = m.querySelector('.pip img');
@@ -719,7 +888,7 @@ function page() {
       clearInterval(tick);
       const label = document.createElement('p');
       label.className = 'stand-in-label';
-      label.textContent = 'inspiration · sketch pending';
+      label.textContent = 'inspiration · comp pending';
       m.appendChild(label);
     };
     const tryLoad = () => {
@@ -758,15 +927,15 @@ function page() {
     if (img.complete && img.naturalWidth === 0 && img.getAttribute('src')) artFailed(img);
     else img.addEventListener('error', () => artFailed(img), { once: true });
   });
-  // A broken inspiration PIP just leaves; nothing depends on it.
-  document.querySelectorAll('.pip img').forEach(img => {
-    const gone = () => img.closest('.pip')?.remove();
+  // A broken inspiration PIP or thumb just leaves; nothing depends on it.
+  document.querySelectorAll('.pip img, .inspo img').forEach(img => {
+    const gone = () => img.closest('.pip, .inspo')?.remove();
     if (img.complete && img.naturalWidth === 0) gone();
     else img.addEventListener('error', gone, { once: true });
   });
 
-  // Inspiration PIP opens the full catalog card in the lightbox.
-  document.querySelectorAll('.pip').forEach(p => p.addEventListener('click', (e) => {
+  // Inspiration PIP or body thumb opens the full catalog card in the lightbox.
+  document.querySelectorAll('.pip, .inspo').forEach(p => p.addEventListener('click', (e) => {
     e.stopPropagation();
     const img = p.querySelector('img');
     if (!img) return;
@@ -813,7 +982,7 @@ function page() {
   const ambient = document.getElementById('ambient');
   document.querySelectorAll('.card').forEach(card => {
     card.addEventListener('mouseenter', () => {
-      const art = card.querySelector('.face.front .media img:not([hidden])') || card.querySelector('.face.front .pip img');
+      const art = card.querySelector('.face.front .media img:not([hidden])') || card.querySelector('.face.front .pip img') || card.querySelector('.face.front .inspo img');
       if (!art || !art.getAttribute('src')) return;
       ambient.style.backgroundImage = 'url("' + art.getAttribute('src') + '")'; ambient.style.opacity = '1';
     });
@@ -861,8 +1030,11 @@ function page() {
   lightbox.addEventListener('click', closeLightbox);
   document.addEventListener('keydown', (e) => { if (e.key === 'Escape' && !lightbox.hidden) closeLightbox(); });
   document.getElementById('canon')?.addEventListener('click', () => answer('canon'));
-  document.getElementById('reroll')?.addEventListener('click', async () => {
-    await fetch('/answer', { method: 'POST', headers: { 'content-type': 'application/json' }, body: JSON.stringify({ optionId: 'reroll', steer: steer() }) });
+  const dealAgain = async (register) => {
+    await fetch('/answer', { method: 'POST', headers: { 'content-type': 'application/json' }, body: JSON.stringify({ optionId: 'reroll', steer: steer(), ...(register ? { register } : {}) }) });
+    await awaitNextRound();
+  };
+  async function awaitNextRound() {
     const grid = document.querySelector('.grid');
     const cardsNow = [...grid.querySelectorAll('.card')];
     const g = grid.getBoundingClientRect();
@@ -879,14 +1051,17 @@ function page() {
     }
     const cardHeight = cardsNow[0] ? cardsNow[0].getBoundingClientRect().height : 0;
     grid.innerHTML = cardsNow.map(() => '<article class="card skeleton"' + (cardHeight ? ' style="height:' + cardHeight + 'px"' : '') + '><div class="card-inner"><div class="face front"><div class="media"><div class="shimmer"></div></div><div class="body"><div class="line tier w40"></div><div class="line title w70"></div><div class="line w90"></div><div class="line w80"></div><div class="line w60"></div><div class="line button"></div></div></div></div></article>').join('');
-    document.getElementById('reroll')?.setAttribute('disabled', '');
+    document.querySelectorAll('.reroll-btn').forEach(b => b.setAttribute('disabled', ''));
     const poll = setInterval(async () => {
       try {
         const status = await (await fetch('/next-status')).json();
         if (status.ready) { clearInterval(poll); location.reload(); }
       } catch { /* server briefly busy */ }
     }, 1200);
-  });
+  }
+  document.getElementById('reroll')?.addEventListener('click', () => dealAgain());
+  document.getElementById('reroll-safer')?.addEventListener('click', () => dealAgain('safer'));
+  document.getElementById('reroll-bolder')?.addEventListener('click', () => dealAgain('bolder'));
 </script>`;
 }
 
@@ -943,22 +1118,29 @@ const server = http.createServer((req, res) => {
       let parsed = {};
       try { parsed = JSON.parse(body); } catch { /* empty steer */ }
       const chosen = options.find((o) => o.id === parsed.optionId);
+      const isReroll = parsed.optionId === 'reroll';
+      // A followup round's pick is not terminal: the table stays open for the
+      // next round (--update), exactly like a re-roll. Detached mode only;
+      // the blocking mode has no update channel, so its picks stay terminal.
+      const followupOpen = Boolean(detachedKey) && payload.followup === true && !isReroll;
       const answer = JSON.stringify({
         optionId: parsed.optionId ?? null,
         steer: parsed.steer ?? '',
+        ...(isReroll && (parsed.register === 'safer' || parsed.register === 'bolder') ? { register: parsed.register } : {}),
+        ...(followupOpen ? { followup: true } : {}),
         ...(chosen?.hero || chosen?.board ? { hero: chosen.hero ?? null, board: chosen.board ?? null } : {}),
         ...(chosen?.sketch ? { sketch: chosen.sketch } : {}),
       });
-      const isReroll = parsed.optionId === 'reroll';
       if (detachedKey) {
         fs.mkdirSync(QUESTION_DIR, { recursive: true });
         fs.writeFileSync(answerFile(detachedKey), answer + '\n');
       } else {
         printAnswer(answer);
       }
-      // A re-roll in detached mode keeps the table open: the client shows a
-      // loading hand and reloads when --update delivers the next round.
-      if (!(isReroll && detachedKey)) setTimeout(() => process.exit(0), 150);
+      // A re-roll or followup pick in detached mode keeps the table open: the
+      // client shows a loading hand and reloads when --update delivers the
+      // next round.
+      if (!((isReroll || followupOpen) && detachedKey)) setTimeout(() => process.exit(0), 150);
     });
     return;
   }
@@ -976,8 +1158,7 @@ server.listen(portArg, '127.0.0.1', () => {
     console.log('Waiting for the user to choose in the browser (Ctrl-C aborts)...');
   }
   if (!hasFlag('no-open')) {
-    const opener = process.platform === 'darwin' ? 'open' : process.platform === 'win32' ? 'start' : 'xdg-open';
-    try { spawn(opener, [url], { stdio: 'ignore', detached: true }).unref(); } catch { /* URL printed anyway */ }
+    openSystemBrowser(url);
   }
   if (timeoutSec > 0) {
     setTimeout(() => {

.cursor/agents/impeccable-asset-producer.md

@@ -14,9 +14,9 @@ Your job is production cleanup, not new art direction. Work only from the approv
 
 Do not redesign. Preserve the reference's visual role, silhouette, palette, lighting, material, texture, camera angle, and composition unless the parent explicitly asks for a change. Preserve perspective only when it belongs to the object or scene itself; if CSS should create the card transform, shadow, rounded clipping, border, or layout, remove that presentation chrome from the raster.
 
-## Decision Sketches
+## Decision Comps
 
-When the parent hands you a decision card packet instead of an approved mock, the job is one sketch: one card, one file, written to the card's declared `sketch` path the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a sketch is reported back, not padded from imagination. Render through the parent's shared frame, including its aspect: the requested surface's first viewport as a flat, matte design sketch in the card's own palette and type character, deliberately unfinished, no photorealism, no gloss; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. The frame is shared across siblings so no sketch looks more finished than another; a finish gap breaks the comparison. The only legible text is the product's real name and one real headline; greek every other text region into indistinct lines, because an invented spec, price, or date in a sketch is a claim PRODUCT.md never made. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a sketch run.
+When the parent hands you a decision card packet instead of an approved mock, the job is one comp: one card, one file, written to the card's declared `sketch` path (the field keeps its wire name) the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a comp is reported back, not padded from imagination. Render the card's direction as a north-star comp at full fidelity: the requested surface's first viewport, prompt led by the surface's own structure (its regions named in order with their scale relationships, never the world's atmosphere), fully committed in the card's own palette, type character, and material world; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. Every sibling renders at the same full fidelity in its own grammar, one surface, one aspect; equal commitment is what keeps the comparison honest. Real product name and real content only; never invent commercial claims, prices, benchmarks, or dates PRODUCT.md does not carry. Write the prompt sidecar beside the file. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a decision-comp run.
 
 ## Input Contract
 

.cursor/agents/impeccable-finish-reviewer.md

@@ -15,12 +15,12 @@ A hard turn ceiling ends the run without warning; a run that ends before the fiv
 
 ## Input Contract
 
-Expect: the original request; the confirmed user answers; the artifact path(s); desktop and mobile screenshot paths captured by the parent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and the approved comp path; and the skill's `reference/craft-floor.md` path. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
+Expect: the original request; the confirmed user answers; the artifact path(s); the screenshots the parent captured, which live in `.impeccable/review/` (on the web, `desktop.png` and `mobile.png`; on native, device-class names such as `phone.png` and `tablet.png`, suffixed per OS on adaptive); a screenshot path the calling brief names is authoritative when the file exists, and `.impeccable/review/` is where to look when the brief names none or a named path is missing, never a filename you invent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and, on a comp-led build, the approved comp path (a code-led build has no approved comp; it passes the chosen decision comp as a separate critique-reference input, labeled as such, and nothing in this file that binds “the approved comp” binds it); and the skill's `reference/craft-floor.md` path. On a native (`ios` / `android` / `adaptive`) build the packet also carries the platform reference path(s) (`reference/ios.md` / `reference/android.md`) and a line saying no detector ran: read the platform reference alongside the craft floor and judge every check in the platform's own conventions, the screenshots are device captures rather than browser viewports, and your floor check is the build's only slop gate. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
 
 ## Checks, in order
 
-1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comps with no recorded pick mean the approval point was skipped, and that is a material finding.
-2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
+1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comp-round comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comp-round comps with no recorded pick mean the approval point was skipped, and that is a material finding. Files under `.impeccable/mocks/decision/` are exempt: they are the direction round's dealt hand, produced before any comp round, and they imply no approval whatever the build path; a code-led build has no comp round at all.
+2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. A critique-reference comp, when one arrived on such a build, is provocation rather than spec: no element matrix, no adaptation citations, no asset obligations; its one contribution is the question of what the image dared that the build did not, and the dares worth adopting enter material_fixes as ordinary ordered fixes. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
 3. **Ceiling.** Against the QUALITY BAR card: name the world's native devices the build left unused, frame, depth, lettering treatment, ornament density, motion. The card governs commitment and finish, never composition.
 4. **Contract, promise by promise.** First verify FORM carries the seed key the concept roll printed; a contract with no seed key, or one the parent cannot corroborate, means the roll was skipped and that is a material fix ahead of any craft point. Then, for each of the five blocks, does the render keep the promise? Apply the memory test to the first viewport.
 5. **Truth.** Demonstration data authored and labeled synthetic; no invented commercial claims; unanswered claims present as marked placeholders, not omissions. Every image-native region of the approved comp shipped as a real asset, not a gradient standing in for one, and every produced asset visibly present in the screenshots; an asset applied at near-zero opacity or buried behind other paint is a compliance token, not a shipped material.
@@ -38,4 +38,4 @@ Return the disposition line first, then exactly five sections: `persistence` (pa
 
 ## Verdict Pass
 
-When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.
+When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent recaptures over the same screenshot files you read in the review round, so re-read those exact paths for this round; a round-stamped filename you invent points at nothing. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.

.cursor/skills/impeccable/SKILL.md

@@ -10,11 +10,11 @@ This skill gives you the tools and permission to create design that earns to be
 Core principles:
 - Go all out. No hedging, no shortcuts. The deliverable must be complete (except assets the user must provide).
 - Dream big and bold. Distinct, beautiful, outstanding and highly inspiring work.
-- Verify in bounded passes, not a loop, and the ceiling covers the whole cycle: screenshots, defect scans, micro-edits, and rebuilds alike. Build fully, inspect once with a batched round (desktop and mobile together), fix everything it shows in one batch, confirm with at most one more round, and stop polishing. Open-ended self-QA burns the user's money doing worse what the finish handoffs do better.
+- Verify in bounded passes, not a loop, and the ceiling covers the whole cycle: screenshots, defect scans, micro-edits, and rebuilds alike. Build fully, inspect once with a batched round (desktop and mobile together on the web; the shipped device classes on a native platform), fix everything it shows in one batch, confirm with at most one more round, and stop polishing. Open-ended self-QA burns the user's money doing worse what the finish handoffs do better.
 
 ## Setup
 
-1. Run `node .cursor/skills/impeccable/scripts/context.mjs` once per session (if the runtime shows this skill's loaded base directory, run `node <skill-base-dir>/scripts/context.mjs`; keep cwd at the user's project). Pass a named source file or route as `--target <path>`. It loads PRODUCT.md, DESIGN.md, the matching surface brief, and native-platform guidance when applicable; follow its directives and do not rerun it.
+1. Run `node <skill-base-dir>/scripts/context.mjs` once per session, where `<skill-base-dir>` is the loaded base directory the runtime reports for this skill; keep cwd at the user's project. That base directory resolves every `node .cursor/skills/impeccable/scripts/...` command in this skill and its references, and `.cursor/skills/impeccable/scripts` is the fallback only when the runtime reports no base directory. Pass a named source file or route as `--target <path>`. It loads PRODUCT.md, DESIGN.md, the matching surface brief, and native-platform guidance when applicable; follow its directives and do not rerun it.
 2. Before acting, load the one playbook that owns the request: the Commands table's reference for an explicit or clearly implied sub-command, or [reference/new-work.md](reference/new-work.md) for a new surface or replacement visual world. Then inspect the target and at least one representative source of incumbent visual truth (tokens, theme, CSS, component, or asset) before editing.
 3. After analysis and direction are resolved, load [reference/craft-floor.md](reference/craft-floor.md) immediately before editing UI. It carries the quality floor, the absolute bans, and the reflexes no detector catches. Do not load it for planning-only work.
 

.cursor/skills/impeccable/reference/android.md

@@ -38,3 +38,9 @@ Would a fluent Android user trust this app, or trip on off-spec components? The
 - **One FAB, one primary action.** Never stack FABs or spend one on a secondary task.
 - **Snackbars for transient feedback** (actionable when useful, never a toast for that); dialogs only for decisions that must interrupt.
 - **Material motion patterns.** Container transform, shared-axis, fade-through, with standard easing and durations; honor the system Remove animations setting with a crossfade or instant cut.
+
+## Verifying the build
+
+- **Screenshots come from the emulator or a connected device, never a browser.** Build and install, then capture with `adb exec-out screencap -p > <path>` (pick a device with `adb -s <serial>` when several are attached). Capture every device class the app ships to, at least one phone and, when tablets are a target, one tablet, and write the files where the review flow expects them.
+- **Dark theme and font scale belong in the pass.** `adb shell cmd uimode night yes` flips the theme; `adb shell settings put system font_scale 1.3` (restore `1.0` after) catches the clipped labels a fixed layout hides; with several targets attached, the capture's `-s <serial>` goes on these commands too.
+- **Emulators give breadth; gestures, refresh rates, and performance need hardware.** Say which one produced the evidence.

.cursor/skills/impeccable/reference/animate.md

@@ -74,12 +74,15 @@ Keep content visible in the default state so failed scripts do not hide the page
 
 Respect autoplay and sound preferences. Any nonessential loop must stop when offscreen or hidden.
 
+Every web animation needs a `prefers-reduced-motion` path with an intentional alternative. Remove or reduce spatial movement while preserving opacity, color, and state transitions that carry meaning. Reduced motion means fewer and gentler animations, not disabling all motion; feedback that confirms an action should remain legible.
+
 ## Verify
 
 - The focal motion is specific to the selected world and surface.
 - Every supporting animation explains feedback, state, or relationship.
 - Interruption and repeated use behave correctly.
 - Desktop, mobile, and keyboard paths remain usable.
+- The `prefers-reduced-motion` path reduces movement without erasing meaningful feedback or state changes.
 - Expensive effects stay smooth on the target device.
 - Removing an animation would lose meaning or authored character, not merely decoration.
 

.cursor/skills/impeccable/reference/bolder.md

@@ -1,5 +1,7 @@
 > **Additional context needed**: which section is the target, and what must stay untouched.
 
+An open direction round owns the word first: "bolder" said while a direction decision is on the table is the Bolder hand register steer, a fresh deal of foreign forms (see new-work.md), not this command. This command refines a surface whose world already shipped.
+
 "Bolder" is an amplification request, and almost always it is scoped to something that already exists. The surrounding page, its system, and its conventions are the given. Your job is to raise one part to the conviction the rest already implies, without rebuilding anything the brief did not name. The reflex answer, reaching for more effects, is the opposite of bold; reject it first.
 
 ## Scope is sovereign

.cursor/skills/impeccable/reference/craft-floor.md

@@ -12,6 +12,7 @@ Each of these is a check on the built result, not an intention. Run them togethe
 - **Type:** body measure 65–75ch, display max 6rem, tracking floor -0.04em, balanced headings, obvious scale and weight steps. Run the real copy at every breakpoint and fix what overflows.
 - **Motion:** one authored moment, not scattered effects and not one identical entrance on every section. Exponential ease-out from an already-visible default. Reach past transform and opacity: blur, backdrop-filter, clip-path, mask, and shadow belong to the palette when they stay smooth.
 - **States:** hover, disabled, loading, error, empty. Plus real content, working controls, responsive composition, keyboard focus.
+- **Browser surfaces:** the parts you did not draw still carry the design. Text selection, the caret, custom scrollbars, focus rings, underline offset, and the numerals in tabular data all ship with browser defaults that belong to no design system. Theme them from the palette. This is the cheapest signal that a page was built rather than assembled, and the one models skip most reliably.
 - **Copy:** the product's own language. Controls name their action; errors name the problem and the recovery.
 - **Coverage:** every brief requirement present and findable within seconds.
 

.cursor/skills/impeccable/reference/degraded/asset-producer.md

@@ -11,9 +11,9 @@ Your job is production cleanup, not new art direction. Work only from the approv
 
 Do not redesign. Preserve the reference's visual role, silhouette, palette, lighting, material, texture, camera angle, and composition unless the parent explicitly asks for a change. Preserve perspective only when it belongs to the object or scene itself; if CSS should create the card transform, shadow, rounded clipping, border, or layout, remove that presentation chrome from the raster.
 
-## Decision Sketches
+## Decision Comps
 
-When the parent hands you a decision card packet instead of an approved mock, the job is one sketch: one card, one file, written to the card's declared `sketch` path the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a sketch is reported back, not padded from imagination. Render through the parent's shared frame, including its aspect: the requested surface's first viewport as a flat, matte design sketch in the card's own palette and type character, deliberately unfinished, no photorealism, no gloss; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. The frame is shared across siblings so no sketch looks more finished than another; a finish gap breaks the comparison. The only legible text is the product's real name and one real headline; greek every other text region into indistinct lines, because an invented spec, price, or date in a sketch is a claim PRODUCT.md never made. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a sketch run.
+When the parent hands you a decision card packet instead of an approved mock, the job is one comp: one card, one file, written to the card's declared `sketch` path (the field keeps its wire name) the moment it renders. The parent runs several of you in parallel, one per card, so your entire contract is this card; generate first, plan never, because the file on disk is the deliverable and the decision page is waiting on it. Work from the card's structured fields and PRODUCT.md alone; a card too thin to brief a comp is reported back, not padded from imagination. Render the card's direction as a north-star comp at full fidelity: the requested surface's first viewport, prompt led by the surface's own structure (its regions named in order with their scale relationships, never the world's atmosphere), fully committed in the card's own palette, type character, and material world; a native app or mobile-first surface is a portrait frame at its device viewport, never a landscape default. Every sibling renders at the same full fidelity in its own grammar, one surface, one aspect; equal commitment is what keeps the comparison honest. Real product name and real content only; never invent commercial claims, prices, benchmarks, or dates PRODUCT.md does not carry. Write the prompt sidecar beside the file. Return one line naming the path and any deviation, nothing more. Everything below this section is the asset-production job; none of it applies to a decision-comp run.
 
 ## Input Contract
 

.cursor/skills/impeccable/reference/degraded/finish-reviewer.md

@@ -11,12 +11,12 @@ A hard turn ceiling ends the run without warning; a run that ends before the fiv
 
 ## Input Contract
 
-Expect: the original request; the confirmed user answers; the artifact path(s); desktop and mobile screenshot paths captured by the parent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and the approved comp path; and the skill's `reference/craft-floor.md` path. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
+Expect: the original request; the confirmed user answers; the artifact path(s); the screenshots the parent captured, which live in `.impeccable/review/` (on the web, `desktop.png` and `mobile.png`; on native, device-class names such as `phone.png` and `tablet.png`, suffixed per OS on adaptive); a screenshot path the calling brief names is authoritative when the file exists, and `.impeccable/review/` is where to look when the brief names none or a named path is missing, never a filename you invent; the direction contract (THESIS, OWN-WORLD, STORY, FIRST VIEWPORT, FORM); PRODUCT.md path; existing hook or detector findings; the chosen world's QUALITY BAR card paths and, on a comp-led build, the approved comp path (a code-led build has no approved comp; it passes the chosen decision comp as a separate critique-reference input, labeled as such, and nothing in this file that binds “the approved comp” binds it); and the skill's `reference/craft-floor.md` path. On a native (`ios` / `android` / `adaptive`) build the packet also carries the platform reference path(s) (`reference/ios.md` / `reference/android.md`) and a line saying no detector ran: read the platform reference alongside the craft floor and judge every check in the platform's own conventions, the screenshots are device captures rather than browser viewports, and your floor check is the build's only slop gate. When the harness can view images, open the screenshots, the comp, and the card first, and inventory the comp's salient elements in your own words before reading the direction contract or any builder-authored summary: a review anchored on the contract inherits whatever the builder's abstraction dropped.
 
 ## Checks, in order
 
-1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comps with no recorded pick mean the approval point was skipped, and that is a material finding.
-2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
+1. **Persistence.** PRODUCT.md exists. When DESIGN.md predates this build (an extension or redesign), it matches the built world; on a new world it is written after this review by the documenter, so its absence here is not a finding. When comp-round comps exist under `.impeccable/mocks/`, an approval record exists too, the surface brief naming the approved comp or an `approved` flag in its sidecar; comp-round comps with no recorded pick mean the approval point was skipped, and that is a material finding. Files under `.impeccable/mocks/decision/` are exempt: they are the direction round's dealt hand, produced before any comp round, and they imply no approval whatever the build path; a code-led build has no comp round at all.
+2. **Fidelity.** Against your own element inventory of the approved comp, never against the contract's summary of it: topology, reading order, focal scale, overlaps and z-order, density, signature geometry, the primary action's treatment (a CTA the comp physically works, dissolves, or stamps is a signature element, and its plain-rectangle rendition is contradicted), navigation items and icons, headline levels and scale relationships. Classify every salient element: match, acceptable adaptation, missing, contradicted, or added without approval. Two rows are mandatory in every matrix. TYPE: the display lettering's character, compression, width, weight, contrast, terminals, against the comp's; a face of a different character is contradicted however the layout matches. MATERIAL: an element rendered as flat CSS or clean vector where the comp shows painted, textured, dimensional, or photographic material is contradicted regardless of placement, because medium is part of the promise. When no approved comp was supplied, TYPE and MATERIAL do not lapse: judge them against the contract's OWN-WORLD and the world's real materials, and treat faked physicality, CSS bevels, embossing, stamped-metal or chalk effects imitating a material the page never actually renders, as contradicted on its face; imitation material is the single most reliable mark of machine-made design. A critique-reference comp, when one arrived on such a build, is provocation rather than spec: no element matrix, no adaptation citations, no asset obligations; its one contribution is the question of what the image dared that the build did not, and the dares worth adopting enter material_fixes as ordinary ordered fixes. An adaptation counts as intentional only when it cites the user answer, surface brief, accessibility need, or product truth that forced it; an uncited deviation is a defect. A missing signature element, a changed topology, or content added without approval fails fidelity and outranks every craft point in material_fixes. When MATERIAL is contradicted on the focal element, or contradiction is the page rather than the exception, stop ordering repairs: make the first material fix a rebuild directive naming the comp regions to re-derive and the assets to produce; a list of patches against a rejected page launders the rejection into an approval. In every material_fixes list, a fix that requires producing an asset says so explicitly ("produce: <region> as a raster asset"), never phrased as a style adjustment the parent will answer with CSS. The comp is the spec for composition, topology, element inventory, density, lettering character, and material; it is not a pixel spec for semantics, accessibility, or responsive reflow, and that allowance covers translation, never replacement.
 3. **Ceiling.** Against the QUALITY BAR card: name the world's native devices the build left unused, frame, depth, lettering treatment, ornament density, motion. The card governs commitment and finish, never composition.
 4. **Contract, promise by promise.** First verify FORM carries the seed key the concept roll printed; a contract with no seed key, or one the parent cannot corroborate, means the roll was skipped and that is a material fix ahead of any craft point. Then, for each of the five blocks, does the render keep the promise? Apply the memory test to the first viewport.
 5. **Truth.** Demonstration data authored and labeled synthetic; no invented commercial claims; unanswered claims present as marked placeholders, not omissions. Every image-native region of the approved comp shipped as a real asset, not a gradient standing in for one, and every produced asset visibly present in the screenshots; an asset applied at near-zero opacity or buried behind other paint is a compliance token, not a shipped material.
@@ -34,4 +34,4 @@ Return the disposition line first, then exactly five sections: `persistence` (pa
 
 ## Verdict Pass
 
-When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.
+When the parent returns with post-fix recaptures, you are scoring, not re-hunting. The parent recaptures over the same screenshot files you read in the review round, so re-read those exact paths for this round; a round-stamped filename you invent points at nothing. The parent's narration of what was fixed is not evidence; a claimed fix you cannot see in the recaptures is unresolved. For each material fix from your review, one line: resolved, partial, or unresolved, tied to what the new screenshots visibly show; a fix answered mechanically, positions moved but the quality the finding named still absent, is partial at best. Then name at most three regressions the fix batch itself introduced, judged by the same matrix rules, and nothing else; no new hunt, no new checks. Return exactly two sections: `verdict` (the scored list) and `remaining` (what stays open, or "clear"), and end with the disposition line recomputed against what remains open; unresolved or partial material findings can never recompute to ship.

.cursor/skills/impeccable/reference/hooks.md

@@ -48,14 +48,20 @@ The first argument is the action. Defaults to `status`.
 5. If `<action>` is `ignore-value`, `ignore-file`, or `ignore-rule`, just print the script output. The default scope is shared `.impeccable/config.json`; add `--local` only when the user explicitly asks for a private exception.
 6. If `<action>` is `status`, just print the script output. Do not add commentary unless the user asked a follow-up question.
 
-## Intentional findings
+## Triage findings
 
-The hook itself never writes ignore config. Persist an exception only after the user explicitly confirms the flagged issue is intentional, and always go through `hook-admin.mjs`.
+The hook itself never writes ignore config; every exception goes through `hook-admin.mjs`. Triage each finding into one of three outcomes:
+
+- **Real design problem**: fix it. Never add an ignore to skip a fix or to push a blocked write through.
+- **Confident false positive or sanctioned exception**: persist the narrowest ignore yourself and disclose it in your reply. The bar is evidence you can name: an intentional demo or fixture, documentation of bad design, literal or domain-appropriate motion (a ball that bounces), or a choice the user already confirmed. Put that evidence in `--reason` as `"<who decided: evidence>"`; write "user confirmed" only when the user actually did.
+- **Unsure**: leave the finding standing and ask the user in one line. Ask once; a one-line question costs less than the hook re-firing on every later edit.
+
+Self-serve stops at `ignore-value`. `ignore-file` and `ignore-rule` silence too much to add on your own judgment; ask the user first.
 
 Prefer the narrowest exception:
 
-- If the finding line shows an exact `ignore-value` command, run that command. This writes shared `.impeccable/config.json` by default.
-- For value-specific findings such as `overused-font` and `bounce-easing`, use `ignore-value` when the user confirms the specific value. Do not use `ignore-rule overused-font` for a specific font.
+- If the finding line shows an `ignore-value <rule> <value>` pair, pass it to `hook-admin.mjs ignore-value` with your `--reason`. This writes shared `.impeccable/config.json` by default.
+- For value-specific findings such as `overused-font` and `bounce-easing`, use `ignore-value` for the specific value. Do not use `ignore-rule overused-font` for a specific font.
 - If the finding has no value-specific command, such as `side-tab`, scope that one rule to the file: `ignore-value <id> "*" --file <path>`. Run `npx impeccable detect <path>` first to see what actually fires there.
 - Reach for `ignore-file <path>` only when the whole file is out of scope for design review: a fixture, a generated artifact, a deliberate slop demo. It silences every rule for that file permanently, including rules that have not been written yet. A real UI surface with one noisy rule wants the file-scoped value ignore above.
 - Use `ignore-rule <id>` only when the user asks to suppress that whole rule across the project. For broad overused-font suppression, use `ignore-rule overused-font --all-values` only when the user asks to ignore overused fonts generally.
@@ -67,10 +73,10 @@ Example value-specific exception:
 node .cursor/skills/impeccable/scripts/hook-admin.mjs ignore-value overused-font Inter --shared --reason "User confirmed Inter is intentional"
 ```
 
-Example intentional motion exception:
+Example self-served exception, with the evidence named:
 
 ```bash
-node .cursor/skills/impeccable/scripts/hook-admin.mjs ignore-value bounce-easing bounce-ball --shared --reason "User confirmed ball bounce animation is intentional"
+node .cursor/skills/impeccable/scripts/hook-admin.mjs ignore-value bounce-easing bounce-ball --shared --reason "Agent: literal ball-bounce animation, bounce easing is the subject"
 ```
 
 Example whole-rule font exception:

.cursor/skills/impeccable/reference/ios.md

@@ -43,3 +43,9 @@ Would a fluent iPhone user trust this app, or pause at off-spec controls? The te
 
 - **System transitions.** Push slides, sheets rise, dismiss reverses the entrance. Custom transitions that fight the navigation model disorient.
 - **Honor Reduce Motion.** Crossfade instead of parallax and large slides.
+
+## Verifying the build
+
+- **Screenshots come from the Simulator, never a browser.** Build and run, then capture with `xcrun simctl io booted screenshot <path>` (with several running, replace `booted` with the target's UDID from `xcrun simctl list devices booted`; display names can collide, the UDID never does). Capture every device class the app ships to, at least one iPhone and, when iPad is a target, one iPad, and write the files where the review flow expects them.
+- **Dark Mode and Dynamic Type belong in the pass.** `xcrun simctl ui booted appearance dark` flips appearance, reusing the capture's UDID when several are booted; a check at a large Dynamic Type size catches the truncation a fixed layout hides.
+- **Simulators give breadth; posture, gestures, and performance need hardware.** Say which one produced the evidence.

.cursor/skills/impeccable/reference/new-work.md

@@ -43,12 +43,14 @@ The script assigns which structure gets built; your top-ranked structure is what
 1. Name the product's unique mechanism in one sentence, the audience's real scene, its cultural home, and what this first surface must prove. Note the page this category always ships and its predictable opposite; name both as the rut and keep them out of the seven-candidate list. A brief that paints its own picture, a product name, a titled artifact, a governing metaphor, adds its literal reading to the rut: spend at most one candidate on it and derive the rest from elsewhere in the audience's world.
 2. From that cultural world, list seven concrete visual systems, artifacts, places, or rituals the audience knows by heart, each with one line on why it resonates and can carry the mechanism, ordered by resonance. The audience's world includes its graphic and screen traditions, not only its physical objects: the notation, publications, identity programs, data graphics, and interfaces it reads daily; a nameable abstract system (a school of poster, a documentation standard) is as concrete a candidate as any artifact. What would this thing look like as a physical object; what did its world look like before the web? Near-duplicates count once. When more than three of the seven share one material family, the derivation stopped at the subject's most obvious artifact; dig until the list spans at least three families.
 3. Turn that material into complete directions: each joins a reusable visual world to a concrete first-surface experience.
-4. Run `node .cursor/skills/impeccable/scripts/concept-seed.mjs --scope direction --mode <mode>` and follow what it prints. This step has no substitute and no skip condition: on a new or replacement world, writing artifact code before this script has run and its assignment is acknowledged is a contract violation, whatever the harness, the model, or the time pressure, because the roll is the mechanism that keeps every run from converging on the category default. The script assigns which direction gets built and deals catalog challengers. Fuse each challenger before judging it: the challenger supplies the form and its system grammar, the product supplies every fact, and clarity wins conflicts. Weigh fused challengers against the assigned direction on exactly two axes, audience identification and product clarity; losing to strong grounded material is a valid outcome, and beating a thin or tool-monoculture list is the point.
-5. Present one direction, fully committed: its world, first viewport, visitor path, signature interaction, cross-surface reach, and honest risk. Alongside it, offer the hand's challengers as named alternates, the weighing's verdict written on each as its one-line case, an honest "fuses poorly because X" included; the weighing informs the user's choice, it never pre-empts it. A hand holds at most three challengers: when the roll deals more, the three strongest join the hand and the rest wait in the re-roll pool, noted in one line; dropping a challenger from the hand itself takes a named product-truth failure, disclosed. Add re-roll with an optional one-line steer. Never present a ranked menu of your own grounded candidates; a lineup of those invites the safest card. The two channels share this structure and differ only in richness: cards and boards on the decision page, names and one-liners through the structured tool; the structured tool's option list also carries the standing exit as its last option.
+4. Run `node .cursor/skills/impeccable/scripts/concept-seed.mjs --scope direction --mode <mode>` and follow what it prints. This step has no substitute and no skip condition: on a new or replacement world, writing artifact code before this script has run and its assignment is acknowledged is a contract violation, whatever the harness, the model, or the time pressure, because the roll is the mechanism that keeps every run from converging on the category default. The script assigns which direction gets built and deals catalog challengers. Fuse each challenger before judging it: the challenger supplies the form and its system grammar, the product supplies every fact, and clarity wins conflicts. Weigh fused challengers against the assigned direction on exactly two axes, audience identification and product clarity; losing to strong grounded material is a valid outcome, and beating a thin or tool-monoculture list is the point. The weighing closes with a verdict per challenger, decided before any borrowing is considered: wins (beats the assigned direction on both axes; it becomes the build candidate), competitive (holds one axis; it stays a full alternate), or declined (loses both). A declined challenger is not spent: name the one discipline of its system the assigned direction lacks, and raise the assigned direction to match before presenting it. A donation transfers ambition and system discipline (a palette's total commitment, a grid's density courage, a form's structural honesty), never the challenger's clothes; a motif lifted from a declined world is a costume note, not a raise, and one world owns the page. Write each raise into the presented direction as its own line, named for its donor; a raise nobody can read did not happen.
+5. Present one direction, fully committed and already raised by the hand it beat, its raises visible as named lines: its world, first viewport, visitor path, signature interaction, cross-surface reach, and honest risk. Alongside it, route each dealt challenger by its verdict: winning and competitive challengers are full alternates carrying their QUALITY BAR cards and one-line case, while declined challengers render demoted, compact and quiet, each carrying its verdict plus what the direction kept from it, never full-size and never silently dropped, each still adoptable on request. The verdict informs the user's choice, it never pre-empts it; the demoted row is the hand's proof of judgment, showing why the dealt worlds made the presented direction better. A hand holds at most three full-card challengers: when the roll deals more, the three strongest join the hand and the rest wait in the re-roll pool, noted in one line; dropping a challenger from the hand itself takes a named product-truth failure, disclosed. Add one card for your own top-ranked grounded candidate when it is not the assigned direction, kicker MY PICK, same anatomy as every card, with an honest risk line naming its familiarity when true: the strongest grounded direction is often the one most runs in this category land on, and the user deciding that trade is the point of showing it. Familiar and effective is a legitimate destination, not a failure of nerve; the pick card and the standing exit serve it at two depths. One pick card, never two, never a ranked list: the rest of your grounded candidates stay yours, because a lineup of them hands selection back to a taste function and invites the safest card. The pick never takes the lead position, and when the dice assign your top candidate there is no pick card; the assigned card notes it also topped your list. Add re-roll with an optional one-line steer, offered in three registers: plain (a fresh hand, same spread), safer (the familiar register: your remaining conventional grounded candidates plus the canon against named competitors), and bolder (foreign forms only, at full commitment). A register is the user's steering on the familiar-to-bold axis, never yours to pre-select; when the answer carries one, re-run the seed with `--register <value>` and the next `--reroll` round, and follow what it prints. A user saying "bolder" or "safer" while a direction round is open means these registers, never the bolder or harden commands. The two channels share this structure and differ only in richness: cards and boards on the decision page, names and one-liners through the structured tool; the structured tool's option list carries the assigned direction, the pick, the winning and competitive challengers, and the standing exit as its last option, while declined challengers fold into the assigned option's description as their kept lines, so the raise survives the text channel too.
 
-The standing exit: every direction round offers one quiet, permanent alternative, the category standard, played straight. It is the user's door, never yours: never recommend it, never weigh it against the roll, never let it soften the dealt directions; the counterweights bind the unchosen default, not the chosen one. When the user takes it, in the canon action, a safer-steer, or plain words asking for the familiar or competitor-like path, convention becomes the commitment: ask once for two or three products this should sit alongside, make their craft level the bar, and execute the canon at full fidelity, without irony or smuggled quirk. A standing preference gets recorded as a brand commitment in PRODUCT.md. Re-roll eliminates every direction already shown, grounded and challenger alike; after two consecutive re-rolls, ask what quality is missing. You may re-roll on your own only on named factual grounds, when the assigned direction cannot carry the product's truth or task; taste is never grounds. The user may re-roll freely, and a user- or brief-pinned direction beats the roll, always. Present the decision visually: write an options payload with the assigned direction leading, the dealt challengers as alternates carrying their QUALITY BAR cards, and re-roll, steer, plus canon enabled; a degraded roll with no challengers still uses the page, as a single text-only card with re-roll. Give every card the same anatomy, thesis, palette, materials, first viewport, honest risk, and the challengers' case lines (run the script with `--schema` for the exact shape); the page renders identity from these fields, and a challenger's catalog image rides as labeled inspiration, never as the promise of the build. Author `canonCard` too: the category standard as one honest card with the same anatomy; the page keeps it subordinate, and the counterweights still bind you. Run `node .cursor/skills/impeccable/scripts/serve-question.mjs --start --payload <file>` (run it with `--schema` first for the exact payload shape). It daemonizes, prints the page URL and a key, and exits immediately; now open that URL for the user, in-app browser first, then the system opener, then showing the URL. Collect the choice with `--wait --key <key>`, repeating while it exits 3; the ANSWER prints as JSON. Exit 4 means the page was closed without an answer: re-present once through the structured question tool, and with no answer there either, proceed unattended with the assigned direction and state the assumptions. A harness that can leave a shell blocked in the background may instead run the script without `--start` and let it auto-open and block. Only a session where no browser can open at all, headless, CI, an eval worker, a remote shell with no display, puts the same decision through the structured question tool instead; the script self-detects these environments and exits 2 with that advice, so treat exit 2 as this fallback, never as an error to retry.
+The standing exit: every direction round offers one quiet, permanent alternative, the category standard, played straight. It is the user's door, never yours: never recommend it, never weigh it against the roll, never let it soften the dealt directions; the counterweights bind the unchosen default, not the chosen one. When the user takes it, in the canon action, a safer-steer, or plain words asking for the familiar or competitor-like path, convention becomes the commitment: ask once for two or three products this should sit alongside, make their craft level the bar, and execute the canon at full fidelity, without irony or smuggled quirk. A standing preference gets recorded as a brand commitment in PRODUCT.md. Re-roll eliminates every direction already shown, grounded and challenger alike; after two consecutive re-rolls, ask what quality is missing. You may re-roll on your own only on named factual grounds, when the assigned direction cannot carry the product's truth or task; taste is never grounds. The user may re-roll freely, and a user- or brief-pinned direction beats the roll, always. Present the decision visually: write an options payload with the assigned direction leading and its raised lines included, the pick card when one exists, the dealt challengers as alternates carrying their QUALITY BAR cards plus each challenger's verdict and kept line, re-roll with its safer and bolder registers, steer, plus canon enabled, and `followup: true` when the execution-contract round will follow (it does whenever image generation exists and no standing build-path preference is recorded); a degraded roll with no challengers still uses the page, as a single text-only card with re-roll. Give every card the same anatomy, thesis, palette, materials, first viewport, honest risk, and the challengers' case lines (run the script with `--schema` for the exact shape); the page renders identity from these fields, routes declined challengers to a demoted row on its own, and a challenger's catalog image rides as labeled inspiration, never as the promise of the build. Author `canonCard` too: the category standard as one honest card with the same anatomy; the page keeps it subordinate, and the counterweights still bind you. Run `node .cursor/skills/impeccable/scripts/serve-question.mjs --start --payload <file>` (run it with `--schema` first for the exact payload shape). It daemonizes, prints the page URL and a key, and exits immediately; now open that URL for the user, in-app browser first, then the system opener, then showing the URL. Collect the choice with `--wait --key <key>`, repeating while it exits 3; the ANSWER prints as JSON. Exit 4 means the page was closed without an answer: re-present once through the structured question tool, and with no answer there either, proceed unattended with the assigned direction and state the assumptions. A harness that can leave a shell blocked in the background may instead run the script without `--start` and let it auto-open and block. Only a session where no browser can open at all, headless, CI, an eval worker, a remote shell with no display, puts the same decision through the structured question tool instead; the script self-detects these environments and exits 2 with that advice, so treat exit 2 as this fallback, never as an error to retry.
 
-When image generation exists, every card also declares a `sketch` path under `.impeccable/sketches/`, the canon card included. Where the harness sandboxes its shell, start the page through the least-sandboxed command path it offers: a sandboxed shell cannot bind the board's port, and the first-attempt failure costs a retry every session. Serve the page first, then produce the sketches; the page shimmer-waits per slot and the user may answer before they land. Render every sketch through one shared frame so the comparison stays about direction, never rendering luck: the requested surface's first viewport as a flat, matte design sketch in that card's own palette and type character, deliberately unfinished, no photorealism, no gloss, identical framing across cards; a candidate whose sketch looks more finished than the others has broken the comparison, not won it. The frame's aspect is the surface's own: a native app or mobile-first surface sketches portrait at its device viewport, a desktop web surface landscape, and the decision page adapts to either, so a phone screen sketched landscape is a broken frame, not a neutral default. The only legible text in a sketch is the product's real name and one real headline; every other text region is greeked, indistinct lines standing where copy will go, because a sketch that renders invented specs, prices, or dates puts claims in front of the user that PRODUCT.md never made. Produce in the order the user reads: the assigned card, then the hand, then canon, each file written the moment it is done. When the harness runs subagents in parallel, fan the set out as one agent per card: each spawn is the shipped asset producer with a single-sketch packet, that card's fields, PRODUCT.md, the shared frame, and the card's declared path, up to four in flight at once. A slot still empty when its agent returns is regenerated inline, and a slot still empty when the user answers is dropped without ceremony; no other supervision is owed. Without parallel subagents, generate in the main thread after serving, in the same reading order, and let the harness's own generation display carry the progress; the wait for the answer follows the last file. A sketch answers which world, never which composition: the comp round still renders its full set, and the chosen card's sketch seeds at most one probe. With no image generation, the cards carry their identity in palette chips and facts, and that page is complete, not a lesser version.
+When image generation exists, every card also declares a `sketch` path under `.impeccable/mocks/decision/` (the field keeps its wire name for compatibility; what it carries is the card's comp), the canon card included. Where the harness sandboxes its shell, start the page through the least-sandboxed command path it offers: a sandboxed shell cannot bind the board's port, and the first-attempt failure costs a retry every session. Serve the page first, then produce the comps; the page shimmer-waits per slot and the user may answer before they land. Each card's image is that direction's north-star comp at full fidelity, produced under the comp discipline in [visualize.md](visualize.md): the requested surface's first viewport, structure-led prompt, real product name and real content, no invented commercial claims, in that card's own palette, type character, and material world, committed all the way. Generation takes the same time at any fidelity, so an unfinished sketch pays sketch quality for comp cost; fairness between cards comes from equal fidelity in each card's own grammar, one surface, one aspect, never from shared unfinishedness. The frame's aspect is the surface's own: a native app or mobile-first surface comps portrait at its device viewport, a desktop web surface landscape, and the decision page adapts to either, so a phone screen comped landscape is a broken frame, not a neutral default. Produce in the order the user reads, the assigned card, then the pick, then the full-card hand, then canon, each file written with its prompt sidecar the moment it is done, so a re-roll's spend front-loads onto the cards read first; declined challengers get no comp, their catalog thumb is their face. When the harness runs subagents in parallel, fan the set out as one agent per card: each spawn is the shipped asset producer with a single-comp packet, that card's fields, PRODUCT.md, the shared frame, and the card's declared path, up to four in flight at once. A slot still empty when its agent returns is regenerated inline, and a slot still empty when the user answers is dropped without ceremony; no other supervision is owed. Without parallel subagents, generate in the main thread after serving, in the same reading order, and let the harness's own generation display carry the progress; the wait for the answer follows the last file. The chosen card's comp is not spent by the choice: on a comp-led build it enters the comp round as compositional option one, and on a code-led build it returns at the finish review as the critique reference, what the image dared that the build did not. The unchosen comps stay in `.impeccable/mocks/decision/` as the round's spent hand; they carry no approval and imply none. With no image generation, the cards carry their identity in palette chips and facts, and that page is complete, not a lesser version; the page then also demotes every challenger's catalog art to a labeled thumbnail on its own, because salience must encode the verdict, never the accident of which cards have images.
+
+The moment the direction lands, one more round on the same open table decides the execution contract. The direction payload declares `followup: true`, so the table stays open after the pick; deliver the build-path payload through `--update` immediately. Two text-only cards. **Comp-led**: a first-viewport comp is generated and it is law, the finish review audits the build against it; boldest composition on the table, fix rounds expected, motion at risk; choosing it makes the comp non-optional, no silent skipping. **Code-led**: no comp of this page and no apology for it; the QUALITY BAR boards still calibrate finish, and the ambition moves into the written contract, the FIRST VIEWPORT block plus a named signature interaction and motion grammar, which the finish reviewer audits in behavior; code-led is not a discount on commitment, the direction still lands fully committed in code. Lead with the chosen world's fit: a costume-heavy catalog world leads comp-led, a quiet or conventional direction leads code-led; the lead is a default, never a decision, and the user flips it freely. A standing preference, voiced once, is recorded as a brand commitment in PRODUCT.md and skips this round on later surfaces. Without image generation there is no fork and no round: code-led is the only path, stated in one line rather than asked. Only a detached table (`--start`) stays open for `--update`: a blocking serve or the structured-tool channel runs the build-path round as its own second question instead, and `followup: true` belongs only on a detached round.
 
 Catalog worlds are working systems, not mood references. When one survives, carry its palette and material, type and composition, topology, controls and state, and responsive rules into the product. When the source is itself an interface language, commit to its native grammar across navigation, content, controls, and states. Open the QUALITY BAR board and hero for the world you build the moment the choice lands, even if you viewed another card earlier; the ANSWER line names the chosen card's images (when the harness only reads files or runs sandboxed, download them into the workspace and open the relative path; sandboxed viewers reject absolute paths outside it). They set the craft level the build must reach, a rendered reference's finish, commitment, and art direction, never the composition; your surface serves this product.
 
@@ -78,7 +80,7 @@ If the work establishes durable strategy for a route or artifact, read its exist
 
 Keep the brief small: scope and visitor mode; audience, job, action/task, proof/content, and constraints; chosen direction and memorable moment; unresolved decisions. Do not copy global product truth or DESIGN.md tokens into it.
 
-Whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports, the locked direction is visualized before it is built, never skipped: load [visualize.md](visualize.md) and follow it, three compositional options rendered and put before the user for approval. This step is proven to produce the most compositional and ambitious work.
+On a comp-led build, whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports, the locked direction is visualized before it is built, never skipped: load [visualize.md](visualize.md) and follow it, three compositional options put before the user for approval, the chosen card's decision comp plus two variations. This step is proven to produce the most compositional and ambitious work. On a code-led build the comp round is skipped by contract, never by drift: the ambition it would have carried lives in the direction contract's FIRST VIEWPORT block and named signature interaction, and the finish reviewer audits those promises in behavior.
 
 For `shape`, return the selected direction to [shape.md](shape.md) and stop before persistence or implementation.
 
@@ -101,8 +103,8 @@ Preserve semantics, accessibility, performance, responsiveness, project conventi
 
 ## 7. Inspect and finish
 
-Inspect desktop and mobile in one batched screenshot round, critique the render against the user's request and the direction contract, fix material gaps, and confirm with one final round; two rounds is the ceiling, and fixes batch between them rather than earning per-tweak screenshots. When an approved comp exists, the critique is a side-by-side: view the comp region and the build region together, the hero and each section as its own crop at legible scale, never one full-page thumbnail, which hides exactly the failures that matter, crude controls, wrong lettering character, flattened material, behind a superficially similar section order. On a Persuade surface, verify the mode did its job: a first-time visitor should know what this is, why it matters, and what to do within seconds, in the form's own vocabulary.
+Inspect the surface's target sizes in one batched screenshot round: desktop and mobile on the web; on a native platform (`ios` / `android` / `adaptive`), the shipped device classes per OS, captured from the simulator or emulator the way the platform reference's Verifying the build section describes. Critique the render against the user's request and the direction contract, fix material gaps, and confirm with one final round; two rounds is the ceiling, and fixes batch between them rather than earning per-tweak screenshots. When an approved comp exists, the critique is a side-by-side: view the comp region and the build region together, the hero and each section as its own crop at legible scale, never one full-page thumbnail, which hides exactly the failures that matter, crude controls, wrong lettering character, flattened material, behind a superficially similar section order. On a Persuade surface, verify the mode did its job: a first-time visitor should know what this is, why it matters, and what to do within seconds, in the form's own vocabulary.
 
-After the second inspection round the build thread's polishing is over: no further defect hunts, micro-edit scripts, or rebuilds here; whatever remains ships through the handoffs, where a fresh context does the finding better and cheaper. Where this harness runs no design hook, run `node .cursor/skills/impeccable/scripts/detect.mjs --json` on the changed targets once here, fix what is mechanical, and pass the remaining findings to the reviewer; a hookless build that skips this ships every tell the hook exists to catch. Capture desktop and mobile screenshots to files, then spawn the shipped finish reviewer, `impeccable-finish-reviewer` (`impeccable_finish_reviewer` in codex; `/impeccable-finish-reviewer` in Cursor; on GitHub Copilot say "Use the impeccable-finish-reviewer agent"), with the original request, confirmed answers, the artifact path, the screenshot paths, its direction contract, existing hook findings, the QUALITY BAR card and approved comp paths, and the craft-floor reference path. The reviewer has no browser; screenshots you fail to pass are checks it cannot run. Never read the shipped agents' definition files before spawning; the harness loads them at spawn, and you owe only the input packet. Wait on any agent with one long timeout rather than a loop of short polls, and spend the wait on the next independent step. Verify its return carries the five contract sections; on an empty or thrashed return, respawn once with the same inputs before doing anything else. This review never runs inside the build thread and never inherits it: spawn the reviewer fresh, with no forked conversation history (`fork_turns: 0` in codex); a reviewer that inherits your transcript inherits your framing, your optimism, and your abstractions, and everything it needs travels in the inputs above. Only a harness whose tool surface has no subagent capability at all substitutes a fresh in-thread pass after stepping fully out of the build context, run from [degraded/finish-reviewer.md](degraded/finish-reviewer.md), and a substituted or failed-and-replaced review is disclosed in one line at finish, never silently. When the reviewer's first material fix is a rebuild directive, fidelity failed wholesale rather than in patches, so skip the fix batch and execute the rebuild immediately: re-derive the named regions, produce the named assets, and send the result back for a verdict, telling the user what is happening rather than asking permission to fix a failure. The user is consulted only when a second rebuild directive arrives, both verdicts on the table, or when rebuilding would discard content the user approved. Otherwise apply the material fixes in one batch, rebuild once, and recapture the same viewports. A recapture measures positions, loading, and overflow; it cannot measure whether a fix reached the quality the finding named, so send the recaptured screenshots back to the same reviewer for a verdict scoring every material fix resolved, partial, or unresolved (through the harness's agent continuation; without one, run the scoring fresh from [degraded/finish-reviewer.md](degraded/finish-reviewer.md)'s Verdict Pass). Fixes scored partial or unresolved get another batch, recapture, and verdict. Two rounds is the budget an unattended run ends at; an attended session's ceiling belongs to the user, so when the second verdict still lists open items, put the table in front of them and let them choose between shipping as it stands and funding another round. Whoever is deciding, stop the moment a round resolves nothing, and the reviewer's findings are the only list you work from, never your own re-opened hunt. Report the final verdict table to the user as it stands, open items included, under the reviewer's own disposition word: a table with open material findings is never announced as a pass, and never under a softer label than the reviewer wrote. Do not run a second detector.
+After the second inspection round the build thread's polishing is over: no further defect hunts, micro-edit scripts, or rebuilds here; whatever remains ships through the handoffs, where a fresh context does the finding better and cheaper. On the web, where this harness runs no design hook, run `node .cursor/skills/impeccable/scripts/detect.mjs --json` on the changed targets once here, fix what is mechanical, and pass the remaining findings to the reviewer; a hookless web build that skips this ships every tell the hook exists to catch. A native platform skips the detector entirely: it reads HTML and CSS and has no verdict on native code, so the reviewer's floor check is the only slop gate and the input packet says so. Capture the screenshots into `.impeccable/review/`, one file per captured viewport (on the web, `desktop.png` and `mobile.png`; on native, one per device class, such as `phone.png` and `tablet.png`, suffixed per OS on adaptive), creating that directory when the harness does not; the paths you pass the reviewer are its spec, and that directory is where it looks when a passed path is missing. Then spawn the shipped finish reviewer, `impeccable-finish-reviewer` (`impeccable_finish_reviewer` in codex; `/impeccable-finish-reviewer` in Cursor; on GitHub Copilot say "Use the impeccable-finish-reviewer agent"), with the original request, confirmed answers, the artifact path, the screenshot paths, its direction contract, existing hook findings, the QUALITY BAR card and approved comp paths (on a code-led build there is no approved comp; the chosen decision comp rides in that slot as the critique reference, named as such), the craft-floor reference path, and on a native platform the platform reference path(s), [ios.md](ios.md) / [android.md](android.md), both on adaptive, plus one line saying no detector ran, so the reviewer judges in the platform's conventions rather than the web's. The reviewer has no browser; screenshots you fail to pass are checks it cannot run. Never read the shipped agents' definition files before spawning; the harness loads them at spawn, and you owe only the input packet. Wait on any agent with one long timeout rather than a loop of short polls, and spend the wait on the next independent step. Verify its return carries the five contract sections; on an empty or thrashed return, respawn once with the same inputs before doing anything else. This review never runs inside the build thread and never inherits it: spawn the reviewer fresh, with no forked conversation history (`fork_turns: 0` in codex); a reviewer that inherits your transcript inherits your framing, your optimism, and your abstractions, and everything it needs travels in the inputs above. Only a harness whose tool surface has no subagent capability at all substitutes a fresh in-thread pass after stepping fully out of the build context, run from [degraded/finish-reviewer.md](degraded/finish-reviewer.md), and a substituted or failed-and-replaced review is disclosed in one line at finish, never silently. When the reviewer's first material fix is a rebuild directive, fidelity failed wholesale rather than in patches, so skip the fix batch and execute the rebuild immediately: re-derive the named regions, produce the named assets, and send the result back for a verdict, telling the user what is happening rather than asking permission to fix a failure. The user is consulted only when a second rebuild directive arrives, both verdicts on the table, or when rebuilding would discard content the user approved. Otherwise apply the material fixes in one batch, rebuild once, and recapture the same viewports over the same files. A recapture measures positions, loading, and overflow; it cannot measure whether a fix reached the quality the finding named, so send the recaptured screenshots back to the same reviewer for a verdict scoring every material fix resolved, partial, or unresolved (through the harness's agent continuation; without one, run the scoring fresh from [degraded/finish-reviewer.md](degraded/finish-reviewer.md)'s Verdict Pass). Fixes scored partial or unresolved get another batch, recapture, and verdict. Two rounds is the budget an unattended run ends at; an attended session's ceiling belongs to the user, so when the second verdict still lists open items, put the table in front of them and let them choose between shipping as it stands and funding another round. Whoever is deciding, stop the moment a round resolves nothing, and the reviewer's findings are the only list you work from, never your own re-opened hunt. Report the final verdict table to the user as it stands, open items included, under the reviewer's own disposition word: a table with open material findings is never announced as a pass, and never under a softer label than the reviewer wrote. Do not run a second detector.
 
 Then spawn the shipped documenter, `impeccable-documenter` (`impeccable_documenter` in codex), with the project root, the artifact path, the direction contract, PRODUCT.md, the [document.md](document.md) reference path, and the boundary to write at; it records DESIGN.md and the sidecar from the built world, ground truth over intention; without subagents the pass runs from [degraded/documenter.md](degraded/documenter.md). A clean detector pass is not finished; finished is the contract kept, the comp honored, the review closed, and the system recorded.

.cursor/skills/impeccable/reference/polish.md

@@ -19,7 +19,7 @@ Fix the cause at the narrowest correct level. Ask when a binding system principl
 
 ## 2. Gather the evidence
 
-Use the feature yourself at representative desktop and mobile sizes. Determine:
+Use the feature yourself at the surface's representative sizes: desktop and mobile on the web; on a native platform (`ios` / `android` / `adaptive`), the shipped device classes on the simulator, emulator, or hardware, captured per the platform reference's Verifying the build section. Determine:
 
 - whether the path is functionally complete;
 - the intended quality bar and time available;
@@ -86,10 +86,10 @@ Do not perfect one corner while leaving the rest below the same quality bar.
 
 Walk the complete path again with mouse, keyboard, and touch where applicable. Check:
 
-- mobile, intermediate, and wide layouts;
+- mobile, intermediate, and wide layouts on the web; phone and tablet size classes in both supported orientations on native;
 - loading, empty, error, success, disabled, long-content, and missing-content states;
 - zoom, contrast, focus, semantics, and screen-reader names;
-- console errors, layout shift, interaction latency, image loading, and supported browsers;
+- console errors, layout shift, interaction latency, and image loading everywhere; supported browsers on the web; supported OS versions, runtime warnings, and dropped frames on native;
 - agreement with DESIGN.md, neighboring features, and the user's scope.
 
 Follow the quality guidance supplied by `context.mjs` and hooks, then run any other relevant QA commands. Context requests a manual scan only when no automatic detector is active; never add another detector pass. Fix real defects and document only narrow intentional exceptions. A clean scan does not replace visual judgment.

.cursor/skills/impeccable/reference/visualize.md

@@ -1,12 +1,12 @@
 # Visualize: Direction Comps & Asset Production
 
-Load this from [new-work.md](new-work.md) whenever any image generation is available, a harness-native tool or the API fallback context.mjs reports. PRODUCT.md and DESIGN.md are preconditions. New-work has already resolved the visual world; this file must not reopen it.
+Load this from [new-work.md](new-work.md) on a comp-led build, when image generation is available (a harness-native tool or the API fallback context.mjs reports). A code-led execution contract skips this file by design, not by drift: its ambition lives in the written direction contract and is audited in behavior, so do not load it for a code-led round. PRODUCT.md and DESIGN.md are preconditions. New-work has already resolved the visual world; this file must not reopen it.
 
 The purpose of a probe is to test composition, narrative, hierarchy, density, focal moment, signature use, and image requirements. It is not a second identity workshop. Keep DESIGN.md's palette, typography direction, material language, component character, imagery stance, and motion grammar fixed.
 
 ## Generate three compositional options
 
-Render three distinct high-fidelity north-star comps of the requested surface, with whatever generation capability exists, saved under `.impeccable/mocks/` so they survive the session. Comp at the surface's own viewport: portrait at device size for a native app or mobile-first surface, desktop landscape otherwise; a phone screen comped landscape misstates the composition before anything gets built against it. Comps are the build thread's own work, never delegated: the thread that writes the comp prompts holds the direction's full context, and it has already seen every comp when the build starts. Open every image you produce or reference by its workspace-relative path, never an absolute one: sandboxed viewers reject absolute paths, and everything under the project root has a relative path. Base them on the real content and the surface concepts already developed with the user. Three is the number: one comp invites rubber-stamping, and the spread between three is what surfaces the composition worth building. A decision-page sketch is not a probe: it chose the direction at deliberately unfinished fidelity, so the three comps render regardless, and the chosen card's sketch seeds at most one of them.
+Render three distinct high-fidelity north-star comps of the requested surface, with whatever generation capability exists, saved under `.impeccable/mocks/` so they survive the session. Comp at the surface's own viewport: portrait at device size for a native app or mobile-first surface, desktop landscape otherwise; a phone screen comped landscape misstates the composition before anything gets built against it. Comps are the build thread's own work, never delegated: the thread that writes the comp prompts holds the direction's full context, and it has already seen every comp when the build starts. Open every image you produce or reference by its workspace-relative path, never an absolute one: sandboxed viewers reject absolute paths, and everything under the project root has a relative path. Base them on the real content and the surface concepts already developed with the user. Three is the number: one comp invites rubber-stamping, and the spread between three is what surfaces the composition worth building. The chosen card's decision comp is the first of the three: it already renders this direction at full fidelity under this file's discipline, so this round generates two more that vary what the first held fixed, and all three go to the approval point together. Only a round that arrives with no decision comp, a degraded roll, an identity-mode page, a direction pinned without the decision round, renders all three here.
 
 - A comp is a designed surface, not a picture of the subject. Lead the generation prompt with the surface's own structure, whatever regions this design actually has, named in order with their scale relationships; a page with no navigation states that instead of inventing one, and an unconventional surface states its unconventional skeleton. A prompt that leads with the world's atmosphere gets a vignette back: the model paints the fish market instead of the fish market's website. Self-check every render: if it could hang as a poster, or reads as a photograph or scene with some text on it, it is not a comp; regenerate with the layout scaffold stated more literally.
 - When the user shortlisted multiple concepts, spread the three across them.
@@ -22,7 +22,7 @@ Show the three together: in the harness when it can display images, otherwise on
 
 Do not begin code until the user approves a direction or explicitly delegates the choice. If they delegate, choose using the task brief, PRODUCT.md, and DESIGN.md, and state the evidence. Approval refines the task concept; it does not modify DESIGN.md.
 
-This approval point has no substitute and no skip condition. When the structured question tool errors, fall back to the decision page; only after both fail may you treat the choice as delegated, and a delegated pick is still recorded exactly as an approval is and disclosed in your first reply, not your last. The finish reviewer treats a build with generated comps and no recorded approval as carrying a material finding.
+This approval point has no substitute and no skip condition. When the structured question tool errors, fall back to the decision page; only after both fail may you treat the choice as delegated, and a delegated pick is still recorded exactly as an approval is and disclosed in your first reply, not your last. The finish reviewer treats a build whose comp round produced comps with no recorded approval as carrying a material finding; decision comps under `.impeccable/mocks/decision/` are the direction round's hand, not comp-round output, and imply no approval on their own.
 
 After approval, record the choice where tools can find it: the approved comp's path goes in the surface brief, and the approved comp's `.json` prompt sidecar gains `"approved": true` (every comp generated through `generate-image.mjs` has one; create it if a native tool didn't). The sidecar travels with the mocks folder, so the approval survives sessions and machines that never see the brief. Then summarize the composition and the parts of the comp that must not be literalized, return to new-work.md, record the direction contract from the approved surface concept, and build.
 

.cursor/skills/impeccable/scripts/concept-seed.mjs

@@ -31,6 +31,16 @@
  *     recomputes what rounds 0..n-1 drew, excludes all of it, and rolls a
  *     fresh assigned index, challengers, and compositions. One base key therefore
  *     reproduces the entire chain of rounds.
+ *   - REGISTER (--register safer|bolder): the user's steering on the
+ *     familiar-to-bold axis, applied to a re-roll round. A register changes
+ *     only what this round instructs, never what it dealt: the same key and
+ *     reroll count reproduce the same deal whatever the register, so the
+ *     exclusion chain never forks. bolder presents the dealt foreign forms
+ *     as the whole hand (first-dealt leads, dice-assigned by deal order);
+ *     safer spends the dealt hand unseen and presents the familiar register,
+ *     the model's conventional grounded candidates plus the canon against
+ *     named competitors, the one sanctioned lineup of the model's own list.
+ *     Registers are user-requested, never pre-selected by the model.
  *   - RATINGS: the reviewer's approval ratings weight the challenger draw
  *     (3-star doubles the odds, 1-star sits out); the approved pool itself
  *     is unchanged.
@@ -41,7 +51,9 @@
  *   node scripts/concept-seed.mjs --scope surface --mode operate --grain flow
  *   node scripts/concept-seed.mjs --scope direction --candidate-count 6
  *   node scripts/concept-seed.mjs --scope direction --mode persuade --from <key> --reroll 1
- *   node scripts/concept-seed.mjs --chosen <challenger-id> --from <key> --scope direction
+ *   node scripts/concept-seed.mjs --scope direction --mode persuade --from <key> --reroll 1 --register bolder
+ *   node scripts/concept-seed.mjs --chosen <challenger-id> --kind challenger --from <key> --scope direction
+ *   node scripts/concept-seed.mjs --kind assigned --from <key> --scope direction
  *
  * --grain names how much of the product is in play: product, flow, view, or
  * region. A docs site, an onboarding flow, a landing page and a data table are
@@ -62,8 +74,13 @@
  * Challenger data resolves in order: a local catalog directory (the private
  * service repo, evals, and tests set IMPECCABLE_CATALOG_DIR), then the roll
  * API at impeccable.style, then a degraded assignment-only seed when both are
- * unavailable. --chosen sends the anonymous choice ping for API-dealt rolls;
- * DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY disables it.
+ * unavailable. The anonymous choice ping fires once per resolved attended
+ * round on API-dealt rolls: --kind names which card class won (assigned,
+ * pick, challenger, canon) so share metrics have a denominator, --chosen
+ * carries the catalog id when a dealt challenger won, and --register rides
+ * along when the round came from a steered hand. Grounded candidates' names
+ * never leave the machine. DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY disables
+ * the ping entirely.
  *
  * Env vars:
  *   IMPECCABLE_CONCEPT_SEED — same as --from; for reproducible eval runs.
@@ -172,17 +189,35 @@ function telemetryDisabled() {
   return Boolean(process.env.IMPECCABLE_NO_TELEMETRY || process.env.DO_NOT_TRACK);
 }
 
-// Anonymous choice ping: records only that a dealt world was selected.
+// Anonymous choice ping: one per resolved attended direction round. kind
+// says which card class won (assigned / pick / challenger / canon), so
+// pick-share and canon-share have a denominator; chosenId rides along only
+// when a dealt catalog world won, and register only when the round came from
+// a steered hand. Grounded candidates' names never leave the machine: they
+// are derived from the user's project, so the ping carries the kind alone.
 // Fire-and-forget; never fails the caller.
-export async function pingChosen({ chosenId, key, scope, mode }) {
-  if (telemetryDisabled() || !chosenId) return false;
+const PING_KINDS = new Set(['assigned', 'pick', 'challenger', 'canon']);
+export async function pingChosen({ chosenId, key, scope, mode, kind, register }) {
+  if (telemetryDisabled()) return false;
+  if (kind && !PING_KINDS.has(kind)) return false;
+  if (register && register !== 'safer' && register !== 'bolder') return false;
+  // Legacy shape: a bare challenger id with no kind stays a valid ping.
+  if (!chosenId && !kind) return false;
+  if ((kind === 'challenger' || !kind) && !chosenId) return false;
   const controller = new AbortController();
   const timer = setTimeout(() => controller.abort(), apiBudgetMs());
   try {
     await fetch(`${API_BASE}/chosen`, {
       method: 'POST',
       headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify({ chosenId, key, scope, mode }),
+      body: JSON.stringify({
+        ...(chosenId ? { chosenId } : {}),
+        key,
+        scope,
+        mode,
+        ...(kind ? { kind } : {}),
+        ...(register ? { register } : {}),
+      }),
       signal: controller.signal,
     });
     return true;
@@ -260,6 +295,7 @@ export function renderConceptSeed({
   scope = 'surface',
   key = process.env.IMPECCABLE_CONCEPT_SEED || crypto.randomBytes(4).toString('hex'),
   reroll = 0,
+  register = null,
   mode = null,
   grain = null,
   platform = null,
@@ -273,6 +309,15 @@ export function renderConceptSeed({
   if (!Number.isInteger(reroll) || reroll < 0) {
     throw new Error('concept-seed: --reroll must be a non-negative integer');
   }
+  if (register !== null && register !== 'safer' && register !== 'bolder') {
+    throw new Error('concept-seed: --register must be safer or bolder');
+  }
+  if (register !== null && reroll < 1) {
+    throw new Error('concept-seed: --register steers a re-roll round; pass --reroll <n> with it');
+  }
+  if (register !== null && scope !== 'direction') {
+    throw new Error('concept-seed: --register applies to direction rounds only');
+  }
   if (mode !== null && !SEED_MODES.has(mode)) {
     throw new Error('concept-seed: --mode must be persuade, operate, read, or experience');
   }
@@ -326,6 +371,7 @@ export function renderConceptSeed({
         scope,
         key,
         reroll,
+        register,
         mode,
         grain,
         platform,
@@ -357,7 +403,11 @@ export function renderConceptSeed({
   survive the current task plus navigation, quiet and dense content,
   interaction and state, and a substantially different future surface. In an
   attended run, present the assigned direction fully committed and offer
-  re-roll; never present a ranked lineup to choose from. Re-roll yourself only
+  re-roll. You may add ONE card for your top-ranked grounded candidate when
+  it is not the assigned direction, kicker MY PICK, with an honest risk line
+  naming its familiarity; one pick card, never a ranked lineup, and the pick
+  never takes the lead position. When the assignment IS your top candidate,
+  there is no pick card. Re-roll yourself only
   on named factual grounds, when the assignment cannot carry the product's
   truth or task; taste is never grounds.`
   : `After ordering the task's grounded structural candidates by resonance,
@@ -374,7 +424,16 @@ export function renderConceptSeed({
   conflicts. Weigh the fused result against the assigned direction on exactly
   two axes, audience identification and product clarity. Losing to strong
   grounded material is a valid outcome; beating a thin or tool-monoculture
-  list is the point. A fused challenger that wins both axes becomes the build.`
+  list is the point. A fused challenger that wins both axes becomes the build.
+  Close the weighing with a verdict per challenger, decided before any
+  borrowing is considered: wins (beats the assigned direction on both axes),
+  competitive (holds one axis), or declined (loses both). A declined
+  challenger is not spent: name the one discipline of its system the assigned
+  direction lacks, and raise the assigned direction to match before
+  presenting it. A donation transfers ambition and system discipline, never
+  the challenger's clothes; one world owns the page. Write each raise as its
+  own named line on the presented direction, and carry every verdict, kept
+  line, and raise into the decision page payload.`
   : `A challenger wins only when its fused result beats the grounded list on
   audience identification and product clarity. It may change task topology or
   interaction, but never the committed visual identity.`;
@@ -399,8 +458,39 @@ Ambitious motion, spatial media, or interaction is welcome when it strengthens
 the product without weakening semantics, performance, or fallback behavior.`;
 
   if (!data) {
-    return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: degraded; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''} --candidate-count ${candidateCount})
-ASSIGNED INDEX: ${buildIndex}
+    // A degraded roll can still serve the safer register, which needs no
+    // catalog at all: the assignment machinery is suppressed entirely, the
+    // same as the non-degraded safer round, because emitting both "the user
+    // picks" and a mandatory numbered build order hands the model two
+    // contradicting instructions and the mandatory one tends to win. The
+    // bolder register is exactly the thing degradation took away, so it
+    // falls back to a plain grounded round, disclosed.
+    const degradedHeader = `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: degraded; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''}${register ? ` --register ${register}` : ''} --candidate-count ${candidateCount})`;
+    if (register === 'safer') {
+      return `${degradedHeader}
+SAFER REGISTER (user-requested): the assigned index is suspended this
+  round; the user picks, and no candidate is mandated. Present the familiar
+  register: your remaining grounded candidates from the conventional end, at
+  most three, as full cards with an honest risk line each, plus the canon
+  executed against two or three named competitors. This is the one sanctioned
+  lineup of your own ranked candidates; it exists only by this explicit
+  request. When the user voices a standing preference for it, record a brand
+  commitment in PRODUCT.md.
+${authorityInstruction}
+A user- or brief-pinned decision beats the roll, always.
+REGISTER (restated for truncated readers): safer, user-requested; the
+assigned index is suspended this round and the user picks; seed key ${key}.
+`;
+    }
+    const degradedRegister = register === 'bolder'
+      ? `BOLDER REGISTER UNAVAILABLE: bolder deals foreign forms, and this roll ran
+  degraded with no catalog and no roll service, so there is nothing bold to
+  deal. Tell the user, then run this round as a plain grounded re-roll; the
+  assignment below applies.
+`
+      : '';
+    return `${degradedHeader}
+${degradedRegister}ASSIGNED INDEX: ${buildIndex}
   ${promotedInstruction}
   The assignment exists to refuse the model's ranking rut, never to outrank
   the user or the brief. Never expose assignment metadata in user-facing labels.
@@ -471,34 +561,76 @@ structure only, never a palette, typeface, or material. Treat them as serious
 rivals to your habitual layout, and keep only what makes this product clearer.${grainNote}\n`
     : '';
   const rerollBlock = reroll > 0
-    ? `RE-ROLL ROUND ${reroll}: every candidate presented in earlier rounds, grounded
-  and challenger alike, is eliminated and may not return reworded. Derive
+    ? `RE-ROLL ROUND ${reroll}${register ? ` (${register.toUpperCase()} REGISTER, user-requested)` : ''}: every candidate presented in earlier rounds, grounded
+  and challenger alike, is eliminated and may not return reworded.${register ? '' : ` Derive
   genuinely new grounded candidates from unexplored angles before judging
-  these fresh challengers.\n`
+  these fresh challengers.`}\n`
     : '';
+  // A register swaps the round's presentation, never its deal: the assigned
+  // index and challenger fetch stay identical so the chain reproduces, and
+  // only the instructions change.
+  const saferBlock = `SAFER REGISTER: the user asked for the familiar end of the spectrum, so this
+  round's dealt hand is spent unseen, stays excluded from future rounds, and
+  is not printed. The assigned index is suspended this round; the user picks. Present the familiar register: your remaining grounded
+  candidates from the conventional end, at most three, as full cards with an
+  honest risk line each, plus the canon executed against two or three named
+  competitors. This is the one sanctioned lineup of your own ranked
+  candidates; it exists only by this explicit request. When the user voices a
+  standing preference for it, record a brand commitment in PRODUCT.md.`;
+  const bolderBlock = `BOLDER REGISTER: the user asked for foreign forms at full commitment, so no
+  grounded direction is presented this round and the assigned index is
+  suspended. The hand is every dealt challenger below, each fused with the
+  product and presented as a full card; the FIRST dealt challenger leads, an
+  assignment by deal order, so the dice still choose. Verdicts and donations
+  apply between the challengers, weighed against the leader. The pick card
+  sits out; the canon stays, as always.`;
   const telemetryBlock = data.source === 'api'
-    ? `TELEMETRY: if the resolved direction uses one of these challengers, rerun
-  this script once with --chosen <challenger-id> --from ${key} --scope ${scope}${mode ? ` --mode ${mode}` : ''}
-  after resolution. The ping is anonymous (chosen id only) and is skipped
-  automatically when DO_NOT_TRACK or IMPECCABLE_NO_TELEMETRY is set.\n`
+    ? `TELEMETRY: after the user's choice resolves, rerun this script once with
+  --kind <assigned|pick|challenger|canon> --from ${key} --scope ${scope}${mode ? ` --mode ${mode}` : ''},
+  adding --chosen <challenger-id> when a dealt challenger won and keeping
+  --register <safer|bolder> when the resolved round came from a steered hand.
+  One ping per resolved attended round. The ping is anonymous, the card kind
+  plus the catalog id when one won; your grounded candidates' names never
+  leave the machine, and the ping is skipped automatically when DO_NOT_TRACK
+  or IMPECCABLE_NO_TELEMETRY is set.\n`
     : '';
-  return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: ${data.source}; approved pool: ${data.poolRevision}; ${data.approvedCount}/${data.catalogCount} human-approved; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''} --candidate-count ${candidateCount} to reproduce this roll against this catalog revision)
-${rerollBlock}ASSIGNED INDEX: ${buildIndex}
+  const assignedBlock = register === null
+    ? `ASSIGNED INDEX: ${buildIndex}
   ${promotedInstruction}
   The assignment exists to refuse the model's ranking rut, never to outrank
-  the user or the brief. Never expose assignment metadata in user-facing labels.
-CHALLENGERS:
+  the user or the brief. Never expose assignment metadata in user-facing labels.`
+    : register === 'safer' ? saferBlock : bolderBlock;
+  // A bolder round has no assigned grounded direction, so the generic
+  // weighing instruction (which measures against the assignment) would
+  // contradict the register; the bolder variant weighs against the leader.
+  const bolderChallengerInstruction = `Fuse each challenger before judging it: the challenger supplies the form
+  and its system grammar, the product supplies every fact, and clarity wins
+  conflicts. Weigh every fused challenger against the fused LEADER, the first
+  dealt, on exactly two axes, audience identification and product clarity;
+  verdicts and donations apply between the challengers, and one that beats
+  the leader on both axes presents as the hand's strongest alternate.`;
+  const roundChallengerInstruction = register === 'bolder' ? bolderChallengerInstruction : challengerInstruction;
+  const challengerSection = register === 'safer'
+    ? ''
+    : `CHALLENGERS:
 ${data.challengers.map(renderChallenger).join('\n')}
-${compositionBlock}${challengerInstruction}
+${compositionBlock}${roundChallengerInstruction}
 When you can view images, open the QUALITY BAR board and hero for any
 challenger you weigh seriously and for the world you build. They exist as a
 craft bar, the finish level and commitment the build is expected to reach,
 never as a mockup to copy; your surface serves this product, not that render.
-${authorityInstruction}
+`;
+  const restated = register === null
+    ? `ASSIGNED INDEX (restated for truncated readers): ${buildIndex}. Build candidate
+${buildIndex} of your own grounded list; seed key ${key}.`
+    : `REGISTER (restated for truncated readers): ${register}, user-requested; the
+assigned index is suspended this round; seed key ${key}.`;
+  return `${scope.toUpperCase()} CONCEPT SEED (key: ${key}; mode: ${mode ?? 'unscoped'}; source: ${data.source}; approved pool: ${data.poolRevision}; ${data.approvedCount}/${data.catalogCount} human-approved; rerun with --scope ${scope}${mode ? ` --mode ${mode}` : ''} --from ${key}${reroll > 0 ? ` --reroll ${reroll}` : ''}${register ? ` --register ${register}` : ''} --candidate-count ${candidateCount} to reproduce this roll against this catalog revision)
+${rerollBlock}${assignedBlock}
+${challengerSection}${authorityInstruction}
 ${richnessInstruction}
 ${telemetryBlock}A user- or brief-pinned decision beats the roll, always.
-ASSIGNED INDEX (restated for truncated readers): ${buildIndex}. Build candidate
-${buildIndex} of your own grounded list; seed key ${key}.
+${restated}
 `;
 }
 
@@ -507,19 +639,25 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
   const fromIdx = args.indexOf('--from');
   const scopeIdx = args.indexOf('--scope');
   const rerollIdx = args.indexOf('--reroll');
+  const registerIdx = args.indexOf('--register');
   const modeIdx = args.indexOf('--mode');
   const grainIdx = args.indexOf('--grain');
   const platformIdx = args.indexOf('--platform');
   const candidateCountIdx = args.indexOf('--candidate-count');
   const chosenIdx = args.indexOf('--chosen');
+  const kindIdx = args.indexOf('--kind');
   try {
-    if (chosenIdx !== -1) {
+    if (chosenIdx !== -1 || kindIdx !== -1) {
       // Choice ping: always exits 0, telemetry must never fail a design flow.
+      // --kind alone pings a non-challenger outcome (assigned/pick/canon);
+      // --chosen alone stays the legacy challenger-win ping.
       const sent = await pingChosen({
-        chosenId: args[chosenIdx + 1],
+        chosenId: chosenIdx !== -1 ? args[chosenIdx + 1] : undefined,
         key: fromIdx !== -1 ? args[fromIdx + 1] : undefined,
         scope: scopeIdx !== -1 ? args[scopeIdx + 1] : undefined,
         mode: modeIdx !== -1 ? args[modeIdx + 1] : undefined,
+        kind: kindIdx !== -1 ? args[kindIdx + 1] : undefined,
+        register: registerIdx !== -1 ? args[registerIdx + 1] : undefined,
       });
       process.stdout.write(sent ? 'choice recorded\n' : 'choice ping skipped\n');
     } else {
@@ -542,6 +680,7 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
           ? args[fromIdx + 1]
           : (process.env.IMPECCABLE_CONCEPT_SEED || crypto.randomBytes(4).toString('hex')),
         reroll: rerollIdx !== -1 ? Number(args[rerollIdx + 1]) : 0,
+        register: registerIdx !== -1 ? args[registerIdx + 1] : null,
         mode: modeIdx !== -1 ? args[modeIdx + 1] : null,
         grain: grainIdx !== -1 ? args[grainIdx + 1] : null,
         platform: platformIdx !== -1 ? args[platformIdx + 1] : null,
@@ -553,6 +692,13 @@ if (process.argv[1] && resolve(process.argv[1]) === fileURLToPath(import.meta.ur
     process.exitCode = 1;
   }
   // A raced-out fetch may still hold a socket; exit explicitly so the CLI
-  // never lingers on a dead network path after output is written.
+  // never lingers on a dead network path after output is written. Destroy
+  // fetch's global undici dispatcher first: process.exit() with a live
+  // keep-alive socket trips a libuv assertion on Windows and aborts the
+  // process after a successful roll (nodejs/node#56645).
+  const dispatcher = globalThis[Symbol.for('undici.globalDispatcher.1')];
+  if (dispatcher && typeof dispatcher.destroy === 'function') {
+    try { await dispatcher.destroy(); } catch { /* exit regardless */ }
+  }
   process.exit(process.exitCode ?? 0);
 }

.cursor/skills/impeccable/scripts/context-signals.mjs

@@ -22,7 +22,7 @@ import path from 'node:path';
 import { fileURLToPath } from 'node:url';
 import { execFileSync } from 'node:child_process';
 import { loadContext, extractPlatform } from './context.mjs';
-import { getCritiqueDir } from './lib/impeccable-paths.mjs';
+import { readLatestSnapshotAcrossTargets } from './critique-storage.mjs';
 
 /** Is there code here at all, or just context files / an empty repo? */
 function hasCode(cwd) {
@@ -34,23 +34,13 @@ function hasCode(cwd) {
 }
 
 /**
- * The most recent critique snapshot across all targets. Filenames are
- * timestamp-prefixed (`<iso>__<slug>.md`), so a lexical sort is chronological.
- * Parses the small frontmatter for score + P0/P1 counts.
+ * Summarize the most recent critique snapshot across all targets.
  */
 function latestCritique(cwd) {
   try {
-    const dir = getCritiqueDir(cwd);
-    if (!fs.existsSync(dir)) return null;
-    const files = fs.readdirSync(dir).filter((f) => f.endsWith('.md')).sort();
-    if (!files.length) return null;
-    const newest = files[files.length - 1];
-    const text = fs.readFileSync(path.join(dir, newest), 'utf-8');
-    const front = text.split('---')[1] || '';
-    const get = (k) => {
-      const m = front.match(new RegExp(`^${k}:\\s*(.+)$`, 'm'));
-      return m ? m[1].trim() : null;
-    };
+    const latest = readLatestSnapshotAcrossTargets({ cwd });
+    if (!latest) return null;
+    const get = (key) => latest.meta[key] ?? null;
     const num = (v) => {
       const n = Number(v);
       return Number.isFinite(n) ? n : null;
@@ -61,7 +51,7 @@ function latestCritique(cwd) {
       p0: num(get('p0')),
       p1: num(get('p1')),
       timestamp: get('timestamp'),
-      file: path.relative(cwd, path.join(dir, newest)),
+      file: path.relative(cwd, latest.path),
     };
   } catch {
     return null;

.cursor/skills/impeccable/scripts/critique-storage.mjs

@@ -105,28 +105,37 @@ function parseFrontmatter(text) {
 }
 
 /**
- * Return all snapshot files for `slug`, sorted oldest → newest.
+ * Return snapshot files matching `suffix`, sorted oldest → newest.
  */
-function listSnapshotsForSlug(slug, cwd) {
+const SNAPSHOT_FILENAME = /^\d{4}-\d{2}-\d{2}T\d{2}-\d{2}-\d{2}Z__.+\.md$/;
+
+function listSnapshots(suffix, cwd) {
   const dir = getCritiqueDir(cwd);
   if (!fs.existsSync(dir)) return [];
-  const suffix = `__${slug}.md`;
   return fs.readdirSync(dir)
-    .filter((f) => f.endsWith(suffix))
+    .filter((f) => SNAPSHOT_FILENAME.test(f) && f.endsWith(suffix))
     .sort()
     .map((f) => path.join(dir, f));
 }
 
+function readLatestSnapshotMatching(suffix, cwd) {
+  const filePath = listSnapshots(suffix, cwd).at(-1);
+  if (!filePath) return null;
+  const body = fs.readFileSync(filePath, 'utf-8');
+  return { path: filePath, body, meta: parseFrontmatter(body) };
+}
+
 /**
  * Return the most recent snapshot for `slug`, or null. Polish reads this
  * to find its fix backlog when the slug matches.
  */
 export function readLatestSnapshot(slug, { cwd = process.cwd() } = {}) {
-  const all = listSnapshotsForSlug(slug, cwd);
-  if (!all.length) return null;
-  const latest = all[all.length - 1];
-  const body = fs.readFileSync(latest, 'utf-8');
-  return { path: latest, body, meta: parseFrontmatter(body) };
+  return readLatestSnapshotMatching(`__${slug}.md`, cwd);
+}
+
+/** Return the most recent snapshot across all targets, or null. */
+export function readLatestSnapshotAcrossTargets({ cwd = process.cwd() } = {}) {
+  return readLatestSnapshotMatching('.md', cwd);
 }
 
 /**
@@ -134,7 +143,7 @@ export function readLatestSnapshot(slug, { cwd = process.cwd() } = {}) {
  * Critique appends a one-line trend to its output using this.
  */
 export function readTrend(slug, { limit = 5, cwd = process.cwd() } = {}) {
-  const all = listSnapshotsForSlug(slug, cwd);
+  const all = listSnapshots(`__${slug}.md`, cwd);
   const slice = all.slice(-limit);
   return slice.map((file) => parseFrontmatter(fs.readFileSync(file, 'utf-8')));
 }

.cursor/skills/impeccable/scripts/detector/browser/injected/index.mjs

@@ -683,6 +683,10 @@ if (IS_BROWSER) {
 
       const reasons = collectVisualContrastReasons(el, style);
       if (reasons.length === 0) continue;
+      // Image-only mode filters here, inside the cap: gradient/opacity/filter
+      // candidates earlier in DOM order must not consume the budget and
+      // starve the url()-backed texts this mode exists to sample.
+      if (options.imageOnly && !reasons.includes('image background')) continue;
 
       const textColor = parseRgb(style.color);
       const fontSize = parseFloat(style.fontSize) || 16;
@@ -1175,6 +1179,7 @@ if (IS_BROWSER) {
   }
 
   async function analyzeVisualContrast(options = {}) {
+    // imageOnly is enforced inside the collector, before the candidate cap.
     const candidates = collectVisualContrastCandidates(options);
     const results = [];
     const shouldScrollOffscreen = options.scrollOffscreen === true;
@@ -1260,9 +1265,16 @@ if (IS_BROWSER) {
 
   function addBrowserFindings(groupMap, el, findings) {
     if (!findings || findings.length === 0) return;
+    // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+    // matching findings for its whole subtree. Applied at this choke point so
+    // every per-element attribution (checks, layout, occlusion, rhythm)
+    // honors it; page-level findings attributed to <body> pass through
+    // untouched, since body has no ignoring ancestor.
+    const kept = findings.filter(f => !scopedIgnoreActive(el, f.type));
+    if (kept.length === 0) return;
     const existing = groupMap.get(el);
-    if (existing) existing.push(...findings);
-    else groupMap.set(el, [...findings]);
+    if (existing) existing.push(...kept);
+    else groupMap.set(el, [...kept]);
   }
 
   function browserFindingsFromMap(groupMap) {
@@ -1620,9 +1632,27 @@ if (IS_BROWSER) {
     for (const node of docClone.querySelectorAll('[id^="impeccable-live-"]')) {
       node.remove();
     }
-    const htmlPatternFindings = checkHtmlPatterns(docClone.outerHTML);
-    if (htmlPatternFindings.length > 0) {
-      const mapped = htmlPatternFindings.map(f => {
+    // Regex findings that name a live selector resolve against the real DOM:
+    // pseudo-element/class segments are stripped (the host element is the
+    // anchor), a selector that matches nothing on this page drops the finding
+    // (the CSS ships here, but the pattern never renders — the live DOM is
+    // ground truth in the browser), and a match under a data-impeccable-ignore
+    // ancestor is waived. Selector-less findings stay page-level.
+    const scopedHtmlFindings = checkHtmlPatterns(docClone.outerHTML).filter(f => {
+      if (!f.selector) return true;
+      const query = String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim().replace(/,\s*(?=,|$)/g, '');
+      if (!query || /^[,\s]*$/.test(query)) return true;
+      let matches;
+      try {
+        matches = document.querySelectorAll(query);
+      } catch {
+        return true;
+      }
+      if (matches.length === 0) return false;
+      return [...matches].some(el => !scopedIgnoreActive(el, f.id));
+    });
+    if (scopedHtmlFindings.length > 0) {
+      const mapped = scopedHtmlFindings.map(f => {
         const item = { type: f.id, detail: f.snippet };
         if (f.severity) {
           item.severity = f.severity;
@@ -1652,8 +1682,27 @@ if (IS_BROWSER) {
     };
   }
 
+  // Visual contrast has three modes. Explicit true runs the full sampled
+  // pass; explicit false disables it entirely (the deterministic-only mode
+  // the test suites use). Unset — the default overlay run — samples ONLY
+  // image-backed text: the one class the analytic walk deliberately skips,
+  // because a url() layer's pixels are unknowable without looking. In-page
+  // sampling draws the source image alone to a canvas (glyph ink never
+  // pollutes it), and a cross-origin image without CORS reports unresolved
+  // instead of guessing.
+  function visualContrastMode(options = {}) {
+    const explicit = typeof options.visualContrast === 'boolean'
+      ? options.visualContrast
+      : typeof window.__IMPECCABLE_CONFIG__?.visualContrast === 'boolean'
+        ? window.__IMPECCABLE_CONFIG__.visualContrast
+        : null;
+    if (explicit === true) return 'full';
+    if (explicit === false) return false;
+    return 'image-only';
+  }
+
   function shouldRunVisualContrast(options = {}) {
-    return options.visualContrast === true || window.__IMPECCABLE_CONFIG__?.visualContrast === true;
+    return visualContrastMode(options) !== false;
   }
 
   function visualContrastOptions(options = {}) {
@@ -1830,6 +1879,7 @@ if (IS_BROWSER) {
       return [];
     }
     const resolvedOptions = visualContrastOptions(options);
+    if (visualContrastMode(options) === 'image-only') resolvedOptions.imageOnly = true;
     const analyses = await analyzeVisualContrast(resolvedOptions);
     if (runtime.generation && runtime.generation !== scanGeneration) return analyses;
     lastVisualContrastAnalyses = analyses;

.cursor/skills/impeccable/scripts/detector/detect-antipatterns-browser.js

@@ -861,6 +861,34 @@ function checkBorders(tag, widths, colors, radius, opts = {}) {
   return findings;
 }
 
+// ─── Scoped ignores: data-impeccable-ignore ─────────────────────────────────
+//
+// An element-scoped waiver that travels with the markup: any element carrying
+// `data-impeccable-ignore="rule-a rule-b"` (or `*`, or an empty value, for
+// every rule) suppresses matching findings from itself and its entire subtree,
+// in every engine that walks elements — the browser overlay, the extension,
+// and the static scan. This is the DOM twin of the line-based
+// `impeccable-disable` comment directives, which the browser cannot apply (a
+// live DOM has no line numbers), and the generalization of the one-off
+// `data-impeccable-allow-kickers` opt-out.
+//
+// The intended use is curated exhibits: a page that documents anti-patterns by
+// example, or renders a deliberate "before" specimen, marks the container once
+// and every engine skips it while still scanning the page around it.
+function scopedIgnoreActive(el, ruleId) {
+  const rule = String(ruleId || '').toLowerCase();
+  let cur = el;
+  while (cur && cur.nodeType === 1) {
+    const attr = typeof cur.getAttribute === 'function' ? cur.getAttribute('data-impeccable-ignore') : null;
+    if (attr != null) {
+      const rules = String(attr).trim().toLowerCase().split(/[\s,]+/).filter(Boolean);
+      if (rules.length === 0 || rules.includes('*') || rules.includes(rule)) return true;
+    }
+    cur = cur.parentElement;
+  }
+  return false;
+}
+
 // Returns true if the given text is composed entirely of emoji characters
 // (plus whitespace / variation selectors). Emojis render as multicolor glyphs
 // regardless of CSS `color`, so contrast checks against the element's text
@@ -1428,6 +1456,26 @@ function cssTextHasDarkRootBg(content, customProps) {
   return false;
 }
 
+// Best-effort extraction of the CSS selector whose declaration block contains
+// the given index in raw CSS text. Lets CSS-text findings carry a live-DOM
+// anchor, so the browser pass can resolve scoped ignores against the actual
+// element and drop patterns that render nowhere on the page. Returns null for
+// @-rule preludes, keyframe steps, nested blocks, and anything that does not
+// read as a selector; those findings stay page-level.
+function enclosingCssSelector(cssText, index) {
+  if (!cssText || !Number.isFinite(index)) return null;
+  const open = cssText.lastIndexOf('{', index);
+  if (open === -1) return null;
+  const prevClose = Math.max(cssText.lastIndexOf('}', open - 1), cssText.lastIndexOf(';', open - 1));
+  const raw = cssText.slice(prevClose + 1, open).trim().replace(/\s+/g, ' ');
+  if (!raw || raw.startsWith('@') || /^\d/.test(raw) || /[{}<]/.test(raw)) return null;
+  // Keyframe steps: percentage steps fail the digit test above, but `from`
+  // and `to` would read as (never-matching) type selectors and get a valid
+  // finding wrongly dropped by the zero-match rule downstream.
+  if (/^(?:from|to)(?:\s*,\s*(?:from|to))*$/i.test(raw)) return null;
+  return raw;
+}
+
 function scanCssTextForGlow(content) {
   const customProps = collectCssCustomProps(content);
   const hasDarkBg = cssTextHasDarkRootBg(content, customProps);
@@ -1739,6 +1787,7 @@ function scanCssTextForPseudoStripe(rawContent) {
       id: 'side-tab',
       snippet: `${selector} — absolute ${thicknessPx}px pseudo-element stripe (${edge}: 0)`,
       index: selectorStart,
+      selector,
     });
   }
   return findings;
@@ -1801,6 +1850,7 @@ function scanCssTextForInsetStripe(content) {
       findings.push({
         id: 'side-tab',
         snippet: `${selector} — inset box-shadow ${ay === 0 ? ax : ay}px stripe (${edge})`,
+        selector,
       });
       break;
     }
@@ -1858,7 +1908,7 @@ function collectMarqueeKeyframes(content) {
 function scanCssTextForMarquee(content, markup = content) {
   const findings = [];
   if (/<marquee\b/i.test(markup)) {
-    findings.push({ id: 'marquee', snippet: '<marquee> element' });
+    findings.push({ id: 'marquee', snippet: '<marquee> element', selector: 'marquee' });
   }
   const marqueeKeyframes = collectMarqueeKeyframes(content);
   if (marqueeKeyframes.size === 0) return findings;
@@ -1873,7 +1923,7 @@ function scanCssTextForMarquee(content, markup = content) {
       const key = `${selector} ${name}`;
       if (seen.has(key)) continue;
       seen.add(key);
-      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"` });
+      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"`, selector });
     }
   }
   return findings;
@@ -2244,8 +2294,10 @@ function checkHtmlPatterns(html, corpora) {
   const purpleHexRe = /#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9|6366f1|764ba2|667eea)\b/gi;
   if (purpleHexRe.test(styleText)) {
     const purpleTextRe = /(?:(?:^|;)\s*color\s*:\s*(?:.*?)(?:#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9))|gradient.*?#(?:7c3aed|8b5cf6|a855f7|764ba2|667eea))/gi;
-    if (purpleTextRe.test(styleText)) {
-      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected' });
+    purpleTextRe.lastIndex = 0;
+    const purpleMatch = purpleTextRe.exec(styleText);
+    if (purpleMatch) {
+      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected', selector: enclosingCssSelector(styleText, purpleMatch.index + 1) || undefined });
     }
   }
 
@@ -2256,7 +2308,7 @@ function checkHtmlPatterns(html, corpora) {
     const start = Math.max(0, gm.index - 200);
     const context = styleText.substring(start, gm.index + gm[0].length + 200);
     if (/gradient/i.test(context)) {
-      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient' });
+      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient', selector: enclosingCssSelector(styleText, gm.index) || undefined });
       break;
     }
   }
@@ -2322,7 +2374,7 @@ function checkHtmlPatterns(html, corpora) {
     const animationToken = bounceMatch[1]
       .split(/[,\s]+/)
       .find((part) => /bounce|elastic|wobble|jiggle|spring/i.test(part));
-    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}` });
+    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}`, selector: enclosingCssSelector(styleText, bounceMatch.index) || undefined });
   }
 
   // Overshoot cubic-bezier
@@ -2331,7 +2383,7 @@ function checkHtmlPatterns(html, corpora) {
   while ((bm = bezierRe.exec(styleText)) !== null) {
     const y1 = parseFloat(bm[2]), y2 = parseFloat(bm[4]);
     if (y1 < -0.1 || y1 > 1.1 || y2 < -0.1 || y2 > 1.1) {
-      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})` });
+      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})`, selector: enclosingCssSelector(styleText, bm.index) || undefined });
       break;
     }
   }
@@ -2364,18 +2416,21 @@ function checkHtmlPatterns(html, corpora) {
 
   const glowHits = scanCssTextForGlow(styleText);
   if (glowHits.length > 0) {
-    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet });
+    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet, selector: enclosingCssSelector(styleText, glowHits[0].index) || undefined });
   }
 
   // Radial-gradient background halo (gradient-drawn sibling of dark-glow)
   const haloHits = scanCssTextForRadialHalo(styleText);
   if (haloHits.length > 0) {
-    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet });
+    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet, selector: enclosingCssSelector(styleText, haloHits[0].index) || undefined });
   }
 
   // --- Generated-UI tells: repeating-gradient stripes ---
-  if (/repeating-(?:linear|radial|conic)-gradient\s*\(/i.test(styleText)) {
-    findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes' });
+  {
+    const stripesMatch = /repeating-(?:linear|radial|conic)-gradient\s*\(/i.exec(styleText);
+    if (stripesMatch) {
+      findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes', selector: enclosingCssSelector(styleText, stripesMatch.index) || undefined });
+    }
   }
 
   // --- Generated-UI tells: two-axis grid-line background ---
@@ -2393,7 +2448,7 @@ function checkHtmlPatterns(html, corpora) {
   // whole gradient layers.
   const gridHits = scanCssTextForGridBackground(styleText);
   if (gridHits.length > 0) {
-    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet });
+    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet, selector: enclosingCssSelector(styleText, gridHits[0].index) || undefined });
   }
 
   // --- Generated-copy tells: "X theater" framing copy ---
@@ -2413,8 +2468,11 @@ function checkHtmlPatterns(html, corpora) {
   // hover:rotate / hover:translate utility on an <img>. Each distinct
   // mechanism is its own finding.
   const imgHoverCss = /\bimg\b[^,{}]*:hover\b[^{}]*\{[^}]*\btransform\s*:\s*(?:scale|rotate|translate|matrix|skew)/i;
-  if (imgHoverCss.test(styleText)) {
-    findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule' });
+  {
+    const imgHoverMatch = imgHoverCss.exec(styleText);
+    if (imgHoverMatch) {
+      findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule', selector: enclosingCssSelector(styleText, imgHoverMatch.index + imgHoverMatch[0].indexOf('{') + 1) || undefined });
+    }
   }
   const imgTagRe = /<img\b[^>]*\bclass\s*=\s*"([^"]*)"/gi;
   let im;
@@ -2461,6 +2519,33 @@ function readOwnBackgroundColor(el, computedStyle) {
   return bg;
 }
 
+// One element's background-color as the cascade walk sees it: computed style
+// first (with the modern-color fallback), then, in static mode only,
+// custom-prop resolution and the inline-shorthand peek. Shared by
+// resolveBackground and resolveGradientStops so both walks read the same
+// surfaces.
+function readCascadeBackgroundColor(current, style, customPropMap) {
+  let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
+  if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
+    // The static engine can return literal "var(--X)" / "oklch(...)" strings.
+    // Resolve through customPropMap so Tailwind v4 color tokens become RGB.
+    if (customPropMap) {
+      bg = parseColorResolved(style.backgroundColor, customPropMap);
+    }
+    if (!bg || bg.a < 0.1) {
+      // Inline-style fallback for colors the static cascade did not surface
+      // on backgroundColor.
+      const rawStyle = current.getAttribute?.('style') || '';
+      const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
+      const inlineBg = bgMatch ? bgMatch[1].trim() : '';
+      if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
+        bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
+      }
+    }
+  }
+  return bg;
+}
+
 function resolveBackground(el, win, customPropMap) {
   let current = el;
   // Translucent layers (0.1 < a < 1) found on the way down to an opaque
@@ -2489,24 +2574,7 @@ function resolveBackground(el, win, customPropMap) {
     // body backgrounds.
     // Real browsers serialize wide-gamut computed values as oklab()/oklch()
     // (e.g. any color-mix() result), which plain parseRgb misses.
-    let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
-    if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
-      // jsdom returns literal "var(--X)" / "oklch(...)" strings. Resolve
-      // through customPropMap so Tailwind v4 color tokens become RGB.
-      if (customPropMap) {
-        bg = parseColorResolved(style.backgroundColor, customPropMap);
-      }
-      if (!bg || bg.a < 0.1) {
-        // Inline-style fallback. jsdom doesn't decompose background
-        // shorthand, so colors set via inline style are otherwise invisible.
-        const rawStyle = current.getAttribute?.('style') || '';
-        const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
-        const inlineBg = bgMatch ? bgMatch[1].trim() : '';
-        if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
-          bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
-        }
-      }
-    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
 
     if (bg && bg.a > 0.1) {
       if (bg.a >= 0.99) return flatten(bg);
@@ -2524,40 +2592,95 @@ function resolveBackground(el, win, customPropMap) {
     //   • on other elements: bail to null and let the caller fall back
     //     to gradient stops (gradient buttons / hero sections are real
     //     bgs worth checking against).
-    if (hasGradientOrUrl) {
-      if (current.tagName === 'BODY' || current.tagName === 'HTML') {
-        return flatten({ r: 255, g: 255, b: 255, a: 1 });
-      }
-      return null;
-    }
+    // A gradient or image with no solid color under it, at any level
+    // including body/html, means the visible ground is that layer itself.
+    // Return null so the caller measures against the actual gradient stops,
+    // or skips when nothing is parseable — skipping beats a wrong ratio.
+    //
+    // Body/html used to assume white here, a guard written for jsdom, which
+    // never decomposed the `background:` shorthand and so could not see the
+    // solid paper color a texture gradient usually sits on. It turned every
+    // light-on-dark page into a wall of low-contrast false positives (a dark
+    // oklch body gradient produced ~120 "on #ffffff" findings on one site).
+    // Both engines can see shorthand solids now — the browser natively, the
+    // static cascade via expandStaticDeclaration + var() resolution — so a
+    // missing solid is real, and the old failure case cannot recur: opaque
+    // stops fully cover any hidden solid (they ARE the ground), alpha stops
+    // composite over the resolved base or the white canvas default, and
+    // unresolvable stops drop rather than guess.
+    if (hasGradientOrUrl) return null;
     current = current.parentElement;
   }
   return flatten({ r: 255, g: 255, b: 255, a: 1 });
 }
 
+// parseGradientColors (shared) reads only the legacy serializations: rgb()
+// and hex stops. Browsers keep modern-space stops in computed backgroundImage
+// exactly as authored — `linear-gradient(oklch(7% 0.006 95), …)` stays oklch —
+// which is what every token-driven page produces. Route those through
+// parseAnyColor so a gradient ground is measurable rather than invisible.
+function parseGradientColorsModern(bgImage) {
+  if (!bgImage || !/gradient/i.test(bgImage)) return [];
+  const colors = parseGradientColors(bgImage);
+  for (const m of bgImage.matchAll(/(?:oklch|oklab|hsla?|hwb)\(\s*[^()]*\)/gi)) {
+    const c = parseAnyColor(m[0]);
+    if (c) colors.push(c);
+  }
+  return colors;
+}
+
 // Walk parents looking for a gradient background and return its color stops.
 // Used as a fallback when resolveBackground() returns null because the
 // effective background is a gradient (no single solid color to compare against).
+// Translucent solid layers found between the element and the gradient (frosted
+// panels, glass washes) are composited over every stop, the same way
+// resolveBackground flattens them over a solid base — raw stops alone would
+// false-flag dark text on a light frosted wash over a dark gradient, and miss
+// the inverse.
 function resolveGradientStops(el, win, customPropMap) {
   let current = el;
+  const overlays = [];
   while (current && current.nodeType === 1) {
     const style = DETECTOR_IS_BROWSER ? getComputedStyle(current) : win.getComputedStyle(current);
     const bgImage = style.backgroundImage || '';
+    // A url() layer anywhere in the stack — alone, or alongside a gradient in
+    // the same declaration (a translucent wash over a texture photo) — paints
+    // pixels the analytic walk cannot know. Measuring the gradient stops over
+    // the wrong base flagged dark ink sitting on a bright gold-leaf image at
+    // 2.6:1; skipping beats a wrong ratio, and the screenshot subsystem owns
+    // image-backed text.
+    if (bgImage && bgImage !== 'none' && /url\s*\(/i.test(bgImage)) return null;
     let stops = null;
     if (bgImage && bgImage !== 'none' && /gradient/i.test(bgImage)) {
-      const parsed = parseGradientColors(bgImage);
+      const parsed = parseGradientColorsModern(bgImage);
       if (parsed.length > 0) stops = parsed;
     }
     if (!stops && !DETECTOR_IS_BROWSER) {
-      // jsdom doesn't decompose `background:` shorthand — peek at the raw inline style
+      // Static mode: peek at the raw inline style for gradients the cascade did not surface
       const rawStyle = current.getAttribute?.('style') || '';
       const bgMatch = rawStyle.match(/background(?:-image)?\s*:\s*([^;]+)/i);
       if (bgMatch && /gradient/i.test(bgMatch[1])) {
-        const parsed = parseGradientColors(bgMatch[1]);
+        const parsed = parseGradientColorsModern(bgMatch[1]);
         if (parsed.length > 0) stops = parsed;
       }
     }
-    if (stops) return compositeGradientStops(stops, current, win, customPropMap);
+    if (stops) {
+      const composited = compositeGradientStops(stops, current, win, customPropMap);
+      if (!composited || overlays.length === 0) return composited;
+      return composited.map(stop => {
+        let acc = stop;
+        for (let i = overlays.length - 1; i >= 0; i--) acc = compositeColorOver(overlays[i], acc);
+        return acc;
+      });
+    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
+    if (bg && bg.a > 0.1) {
+      // An opaque surface above the gradient means the gradient never shows
+      // through here; resolveBackground would have returned it, so reaching
+      // this is defensive — bail rather than measure the wrong layer.
+      if (bg.a >= 0.99) return null;
+      overlays.push(bg);
+    }
     current = current.parentElement;
   }
   return null;
@@ -2777,6 +2900,10 @@ function checkElementColorsDOM(el) {
   const rect = el.getBoundingClientRect();
   if (rect.width < 10 || rect.height < 10) return [];
   const style = getComputedStyle(el);
+  // Invisible at rest: hidden scene variants (opacity-0 carousels, swap
+  // decks) are not user-visible, and measuring their inherited colors against
+  // whatever surface happens to sit behind the stack is noise, not audit.
+  if (style.visibility === 'hidden' || effectiveOpacityDOM(el) <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
   let effectiveBg = resolveBackground(el);
@@ -3589,11 +3716,13 @@ function checkElementGlowDOM(el) {
   // If resolveBackground returns null (gradient), try to infer from the gradient colors
   let parentBg = el.parentElement ? resolveBackground(el.parentElement) : resolveBackground(el);
   if (!parentBg) {
-    // Gradient background — sample its colors to determine if it's dark
+    // Gradient background — sample its colors to determine if it's dark.
+    // Modern-syntax parsing matters here: body-level gradients now reach this
+    // fallback in browser mode, and their stops usually serialize as oklch.
     let cur = el.parentElement;
     while (cur && cur.nodeType === 1) {
       const bgImage = getComputedStyle(cur).backgroundImage || '';
-      const gradColors = parseGradientColors(bgImage);
+      const gradColors = parseGradientColorsModern(bgImage);
       if (gradColors.length > 0) {
         // Average the gradient colors
         const avg = { r: 0, g: 0, b: 0 };
@@ -4435,6 +4564,14 @@ function checkElementBorders(tag, style, overrides, resolvedRadius, el = null) {
 }
 
 function checkElementColors(el, style, tag, window, customPropMap, hasAnchorInheritRule) {
+  // Invisible at rest, static twin of the browser walk's skip: opacity does
+  // not inherit, so walk ancestors multiplying declared opacity down.
+  if (style.visibility === 'hidden') return [];
+  let effOpacity = 1;
+  for (let cur = el; cur && cur.nodeType === 1 && effOpacity > 0.02; cur = cur.parentElement) {
+    effOpacity *= parseFloat(window.getComputedStyle(cur).opacity || '1');
+  }
+  if (effOpacity <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
 
@@ -5593,6 +5730,11 @@ function isRenderedForBrowserRule(el) {
 function checkElementTextOverflowDOM(el) {
   const tag = el.tagName.toLowerCase();
   if (TEXT_OVERFLOW_SKIP_TAGS.has(tag)) return [];
+  // scrollWidth/clientWidth are CSS box-model metrics; on SVG content Chrome
+  // returns arbitrary non-zero values for both (a <text> reported 78/48 while
+  // its rendered length sat comfortably inside its box), so the delta is
+  // noise, not overflow. SVG clips to its own viewport anyway.
+  if (el.namespaceURI === 'http://www.w3.org/2000/svg') return [];
   if (!isRenderedForBrowserRule(el)) return [];
   // Only the element that actually owns overflowing text — not its ancestors,
   // which inherit a wider scrollWidth from the spilling descendant.
@@ -5977,6 +6119,22 @@ function isPaintedForOcclusion(el) {
 // path is pure geometry and runs anywhere on the page.
 const OCCLUSION_TEXT_SKIP_TAGS = new Set(['script', 'style', 'noscript', 'template', 'title']);
 
+// An element whose effective opacity multiplies out to ~0 paints nothing at
+// rest: it is not user-visible, so visual findings on it (contrast, occlusion)
+// measure a state nobody sees. Browser-only — the walk needs live computed
+// styles. Cycling scenes that fade such elements in later are the screenshot
+// subsystem's territory, not the analytic walk's.
+function effectiveOpacityDOM(el) {
+  let o = 1;
+  // Walk all the way through body and html: `body { opacity: 0 }` page-fade
+  // wrappers hide every descendant just as thoroughly as a local wrapper.
+  for (let cur = el; cur && cur.nodeType === 1; cur = cur.parentElement) {
+    o *= parseFloat(getComputedStyle(cur).opacity || '1');
+    if (o <= 0.02) return 0;
+  }
+  return o;
+}
+
 function checkTextOcclusionDOM() {
   const findings = [];
   const seenVictims = new Set();
@@ -6004,6 +6162,11 @@ function checkTextOcclusionDOM() {
     }
     return false;
   };
+  // The classic occluder shape this rules out is an opacity-0 interaction
+  // layer — a range scrubber stretched over a before/after comparison — which
+  // elementFromPoint still returns and whose UA background-color otherwise
+  // reads as an opaque box.
+  const effectiveOpacity = effectiveOpacityDOM;
 
   // Collect renderable text owners in / near the first viewport for the
   // elementFromPoint probe. SVG <text> counts too.
@@ -6016,6 +6179,7 @@ function checkTextOcclusionDOM() {
     const text = inSvg ? (el.textContent || '').trim() : elementDirectText(el);
     if (text.length < 2) continue;
     if (!isPaintedForOcclusion(el)) continue;
+    if (effectiveOpacity(el) <= 0.02) continue;
     let rect; try { rect = el.getBoundingClientRect(); } catch { continue; }
     if (rect.width < 6 || rect.height < 6) continue;
     // Viewport-bound probe: keep text whose box overlaps the live viewport.
@@ -6049,6 +6213,7 @@ function checkTextOcclusionDOM() {
         if (top === el || el.contains(top) || top.contains(el)) continue;
         const topCs = getComputedStyle(top);
         if (isFloated(topCs) || isMarqueeish(top, topCs) || isPinnedOverlay(top)) continue;
+        if (effectiveOpacity(top) <= 0.02) continue;
         const topTag = top.tagName.toLowerCase();
         // Text sitting under a raw image/video is contrast territory (deduped
         // against the pixel low-contrast rule); leave those alone here.
@@ -6941,6 +7106,10 @@ if (IS_BROWSER) {
 
       const reasons = collectVisualContrastReasons(el, style);
       if (reasons.length === 0) continue;
+      // Image-only mode filters here, inside the cap: gradient/opacity/filter
+      // candidates earlier in DOM order must not consume the budget and
+      // starve the url()-backed texts this mode exists to sample.
+      if (options.imageOnly && !reasons.includes('image background')) continue;
 
       const textColor = parseRgb(style.color);
       const fontSize = parseFloat(style.fontSize) || 16;
@@ -7433,6 +7602,7 @@ if (IS_BROWSER) {
   }
 
   async function analyzeVisualContrast(options = {}) {
+    // imageOnly is enforced inside the collector, before the candidate cap.
     const candidates = collectVisualContrastCandidates(options);
     const results = [];
     const shouldScrollOffscreen = options.scrollOffscreen === true;
@@ -7518,9 +7688,16 @@ if (IS_BROWSER) {
 
   function addBrowserFindings(groupMap, el, findings) {
     if (!findings || findings.length === 0) return;
+    // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+    // matching findings for its whole subtree. Applied at this choke point so
+    // every per-element attribution (checks, layout, occlusion, rhythm)
+    // honors it; page-level findings attributed to <body> pass through
+    // untouched, since body has no ignoring ancestor.
+    const kept = findings.filter(f => !scopedIgnoreActive(el, f.type));
+    if (kept.length === 0) return;
     const existing = groupMap.get(el);
-    if (existing) existing.push(...findings);
-    else groupMap.set(el, [...findings]);
+    if (existing) existing.push(...kept);
+    else groupMap.set(el, [...kept]);
   }
 
   function browserFindingsFromMap(groupMap) {
@@ -7878,9 +8055,27 @@ if (IS_BROWSER) {
     for (const node of docClone.querySelectorAll('[id^="impeccable-live-"]')) {
       node.remove();
     }
-    const htmlPatternFindings = checkHtmlPatterns(docClone.outerHTML);
-    if (htmlPatternFindings.length > 0) {
-      const mapped = htmlPatternFindings.map(f => {
+    // Regex findings that name a live selector resolve against the real DOM:
+    // pseudo-element/class segments are stripped (the host element is the
+    // anchor), a selector that matches nothing on this page drops the finding
+    // (the CSS ships here, but the pattern never renders — the live DOM is
+    // ground truth in the browser), and a match under a data-impeccable-ignore
+    // ancestor is waived. Selector-less findings stay page-level.
+    const scopedHtmlFindings = checkHtmlPatterns(docClone.outerHTML).filter(f => {
+      if (!f.selector) return true;
+      const query = String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim().replace(/,\s*(?=,|$)/g, '');
+      if (!query || /^[,\s]*$/.test(query)) return true;
+      let matches;
+      try {
+        matches = document.querySelectorAll(query);
+      } catch {
+        return true;
+      }
+      if (matches.length === 0) return false;
+      return [...matches].some(el => !scopedIgnoreActive(el, f.id));
+    });
+    if (scopedHtmlFindings.length > 0) {
+      const mapped = scopedHtmlFindings.map(f => {
         const item = { type: f.id, detail: f.snippet };
         if (f.severity) {
           item.severity = f.severity;
@@ -7910,8 +8105,27 @@ if (IS_BROWSER) {
     };
   }
 
+  // Visual contrast has three modes. Explicit true runs the full sampled
+  // pass; explicit false disables it entirely (the deterministic-only mode
+  // the test suites use). Unset — the default overlay run — samples ONLY
+  // image-backed text: the one class the analytic walk deliberately skips,
+  // because a url() layer's pixels are unknowable without looking. In-page
+  // sampling draws the source image alone to a canvas (glyph ink never
+  // pollutes it), and a cross-origin image without CORS reports unresolved
+  // instead of guessing.
+  function visualContrastMode(options = {}) {
+    const explicit = typeof options.visualContrast === 'boolean'
+      ? options.visualContrast
+      : typeof window.__IMPECCABLE_CONFIG__?.visualContrast === 'boolean'
+        ? window.__IMPECCABLE_CONFIG__.visualContrast
+        : null;
+    if (explicit === true) return 'full';
+    if (explicit === false) return false;
+    return 'image-only';
+  }
+
   function shouldRunVisualContrast(options = {}) {
-    return options.visualContrast === true || window.__IMPECCABLE_CONFIG__?.visualContrast === true;
+    return visualContrastMode(options) !== false;
   }
 
   function visualContrastOptions(options = {}) {
@@ -8088,6 +8302,7 @@ if (IS_BROWSER) {
       return [];
     }
     const resolvedOptions = visualContrastOptions(options);
+    if (visualContrastMode(options) === 'image-only') resolvedOptions.imageOnly = true;
     const analyses = await analyzeVisualContrast(resolvedOptions);
     if (runtime.generation && runtime.generation !== scanGeneration) return analyses;
     lastVisualContrastAnalyses = analyses;

.cursor/skills/impeccable/scripts/detector/detect-antipatterns.mjs

@@ -35,6 +35,7 @@ export { detectUrl, createBrowserDetector } from './engines/browser/detect-url.m
 export { detectText, extractStyleBlocks, extractCSSinJS } from './engines/regex/detect-text.mjs';
 export {
   walkDir,
+  hasScannableExtension,
   SCANNABLE_EXTENSIONS,
   SKIP_DIRS,
   buildImportGraph,

.cursor/skills/impeccable/scripts/detector/engines/static-html/css-cascade.mjs

@@ -226,6 +226,10 @@ const STATIC_INHERITED_PROPS = new Set([
   'color', 'fontFamily', 'fontSize', 'fontStyle', 'fontWeight', 'fontVariant',
   'lineHeight', 'letterSpacing', 'textTransform', 'textAlign', 'hyphens',
   'webkitHyphens',
+  // visibility inherits in real CSS, and the invisible-at-rest contrast skip
+  // relies on descendants of a hidden container computing as hidden. A child
+  // that declares `visibility: visible` still overrides the inherited value.
+  'visibility',
 ]);
 
 const STATIC_DEFAULT_STYLE = {
@@ -278,6 +282,7 @@ const STATIC_DEFAULT_STYLE = {
   marginLeft: '0px',
   position: 'static',
   visibility: 'visible',
+  opacity: '1',
   top: 'auto',
   right: 'auto',
   bottom: 'auto',
@@ -334,6 +339,7 @@ const STATIC_PROP_MAP = {
   'margin-left': 'marginLeft',
   'position': 'position',
   'visibility': 'visibility',
+  'opacity': 'opacity',
   'top': 'top',
   'right': 'right',
   'bottom': 'bottom',

.cursor/skills/impeccable/scripts/detector/engines/static-html/detect-html.mjs

@@ -28,6 +28,7 @@ import {
   checkCreamPalette,
   checkHtmlPatterns,
   checkKickerAboveHeadingFromDoc,
+  scopedIgnoreActive,
   checkNumberedSectionLabelsFromDoc,
   checkPageLayout,
   checkPageQualityFromDoc,
@@ -138,10 +139,21 @@ async function detectHtml(filePath, options = {}) {
         domutils,
       };
     });
-  } catch {
-    return detectText(html, filePath, options);
+  } catch (err) {
+  if (!globalThis.__impeccableStaticHtmlWarned) {
+    globalThis.__impeccableStaticHtmlWarned = true;
+
+    process.stderr.write(
+  'impeccable detect: DEGRADED - HTML parser modules unavailable ' +
+  '(htmlparser2, css-select, css-tree, domutils).\n' +
+  'Falling back to regex matching. Custom properties, selector matching and computed ' +
+  'contrast are NOT evaluated; findings are an undercount, not a clean bill of health.\n'
+);
   }
 
+  return detectText(html, filePath, options);
+}
+
   const resolvedPath = path.resolve(filePath);
   const fileDir = path.dirname(resolvedPath);
   const root = profileStep(profile, {
@@ -171,6 +183,9 @@ async function detectHtml(filePath, options = {}) {
       const tag = el.tagName.toLowerCase();
       const style = window.getComputedStyle(el);
       for (const f of runElementCheck(rule.id, () => rule.run(el, tag, style, window, customPropMap))) {
+        // Element-scoped waivers: a data-impeccable-ignore ancestor suppresses
+        // matching findings for its subtree, same as the browser walk.
+        if (scopedIgnoreActive(el, f.id)) continue;
         findings.push(finding(f.id, filePath, f.snippet));
       }
     }
@@ -238,6 +253,17 @@ async function detectHtml(filePath, options = {}) {
     for (const f of runPageCheck('html-patterns', () => checkHtmlPatterns(html, patternCorpora).filter(item =>
       item.id !== 'bounce-easing' && item.id !== 'layout-transition'
     ))) {
+      // Selector-backed page findings honor scoped waivers here too, matching
+      // the browser pass: resolve the selector and drop the finding when an
+      // ignoring ancestor covers a match. Unlike the browser, an unmatched
+      // selector keeps the finding — static scans see partial documents.
+      if (f.selector) {
+        let matches = null;
+        try {
+          matches = document.querySelectorAll(String(f.selector).replace(/::?[a-zA-Z-]+(\([^)]*\))?/g, '').trim());
+        } catch { matches = null; }
+        if (matches && matches.length > 0 && [...matches].every(el => scopedIgnoreActive(el, f.id))) continue;
+      }
       const item = finding(f.id, filePath, f.snippet);
       // Position-aware severity promotion: checks may attach a per-finding
       // severity (e.g. a pulsing dot inside a header/nav landmark) that

.cursor/skills/impeccable/scripts/detector/node/file-system.mjs

@@ -26,11 +26,20 @@ const HIDDEN_SOURCE_DIRS = new Set(['.vitepress', '.vuepress', '.storybook']);
 const SCANNABLE_EXTENSIONS = new Set([
   '.html', '.htm', '.css', '.scss', '.sass', '.less',
   '.jsx', '.tsx', '.js', '.ts',
-  '.vue', '.svelte', '.astro',
+  '.vue', '.svelte', '.astro', '.blade.php',
 ]);
 
 const HTML_EXTENSIONS = new Set(['.html', '.htm']);
 
+function hasScannableExtension(filename) {
+  const lower = filename.toLowerCase();
+  if (SCANNABLE_EXTENSIONS.has(path.extname(lower))) return true;
+  for (const ext of SCANNABLE_EXTENSIONS) {
+    if (ext.indexOf('.', 1) !== -1 && lower.endsWith(ext)) return true;
+  }
+  return false;
+}
+
 const IMPORT_SPECIFIER_PATTERNS = [
   /import\s+(?:[\s\S]*?from\s+)?['"]([^'"]+)['"]/g,
   /@import\s+(?:url\(\s*)?['"]?([^'");\s]+)['"]?\s*\)?/g,
@@ -46,7 +55,7 @@ function walkDir(dir) {
     if (entry.isDirectory() && entry.name.startsWith('.') && !HIDDEN_SOURCE_DIRS.has(entry.name)) continue;
     const full = path.join(dir, entry.name);
     if (entry.isDirectory()) files.push(...walkDir(full));
-    else if (SCANNABLE_EXTENSIONS.has(path.extname(entry.name).toLowerCase())) files.push(full);
+    else if (hasScannableExtension(entry.name)) files.push(full);
   }
   return files;
 }
@@ -194,6 +203,7 @@ export {
   SKIP_DIRS,
   SCANNABLE_EXTENSIONS,
   HTML_EXTENSIONS,
+  hasScannableExtension,
   walkDir,
   resolveImport,
   buildImportGraph,

.cursor/skills/impeccable/scripts/detector/rules/checks.mjs

@@ -70,6 +70,34 @@ function checkBorders(tag, widths, colors, radius, opts = {}) {
   return findings;
 }
 
+// ─── Scoped ignores: data-impeccable-ignore ─────────────────────────────────
+//
+// An element-scoped waiver that travels with the markup: any element carrying
+// `data-impeccable-ignore="rule-a rule-b"` (or `*`, or an empty value, for
+// every rule) suppresses matching findings from itself and its entire subtree,
+// in every engine that walks elements — the browser overlay, the extension,
+// and the static scan. This is the DOM twin of the line-based
+// `impeccable-disable` comment directives, which the browser cannot apply (a
+// live DOM has no line numbers), and the generalization of the one-off
+// `data-impeccable-allow-kickers` opt-out.
+//
+// The intended use is curated exhibits: a page that documents anti-patterns by
+// example, or renders a deliberate "before" specimen, marks the container once
+// and every engine skips it while still scanning the page around it.
+function scopedIgnoreActive(el, ruleId) {
+  const rule = String(ruleId || '').toLowerCase();
+  let cur = el;
+  while (cur && cur.nodeType === 1) {
+    const attr = typeof cur.getAttribute === 'function' ? cur.getAttribute('data-impeccable-ignore') : null;
+    if (attr != null) {
+      const rules = String(attr).trim().toLowerCase().split(/[\s,]+/).filter(Boolean);
+      if (rules.length === 0 || rules.includes('*') || rules.includes(rule)) return true;
+    }
+    cur = cur.parentElement;
+  }
+  return false;
+}
+
 // Returns true if the given text is composed entirely of emoji characters
 // (plus whitespace / variation selectors). Emojis render as multicolor glyphs
 // regardless of CSS `color`, so contrast checks against the element's text
@@ -637,6 +665,26 @@ function cssTextHasDarkRootBg(content, customProps) {
   return false;
 }
 
+// Best-effort extraction of the CSS selector whose declaration block contains
+// the given index in raw CSS text. Lets CSS-text findings carry a live-DOM
+// anchor, so the browser pass can resolve scoped ignores against the actual
+// element and drop patterns that render nowhere on the page. Returns null for
+// @-rule preludes, keyframe steps, nested blocks, and anything that does not
+// read as a selector; those findings stay page-level.
+function enclosingCssSelector(cssText, index) {
+  if (!cssText || !Number.isFinite(index)) return null;
+  const open = cssText.lastIndexOf('{', index);
+  if (open === -1) return null;
+  const prevClose = Math.max(cssText.lastIndexOf('}', open - 1), cssText.lastIndexOf(';', open - 1));
+  const raw = cssText.slice(prevClose + 1, open).trim().replace(/\s+/g, ' ');
+  if (!raw || raw.startsWith('@') || /^\d/.test(raw) || /[{}<]/.test(raw)) return null;
+  // Keyframe steps: percentage steps fail the digit test above, but `from`
+  // and `to` would read as (never-matching) type selectors and get a valid
+  // finding wrongly dropped by the zero-match rule downstream.
+  if (/^(?:from|to)(?:\s*,\s*(?:from|to))*$/i.test(raw)) return null;
+  return raw;
+}
+
 function scanCssTextForGlow(content) {
   const customProps = collectCssCustomProps(content);
   const hasDarkBg = cssTextHasDarkRootBg(content, customProps);
@@ -948,6 +996,7 @@ function scanCssTextForPseudoStripe(rawContent) {
       id: 'side-tab',
       snippet: `${selector} — absolute ${thicknessPx}px pseudo-element stripe (${edge}: 0)`,
       index: selectorStart,
+      selector,
     });
   }
   return findings;
@@ -1010,6 +1059,7 @@ function scanCssTextForInsetStripe(content) {
       findings.push({
         id: 'side-tab',
         snippet: `${selector} — inset box-shadow ${ay === 0 ? ax : ay}px stripe (${edge})`,
+        selector,
       });
       break;
     }
@@ -1067,7 +1117,7 @@ function collectMarqueeKeyframes(content) {
 function scanCssTextForMarquee(content, markup = content) {
   const findings = [];
   if (/<marquee\b/i.test(markup)) {
-    findings.push({ id: 'marquee', snippet: '<marquee> element' });
+    findings.push({ id: 'marquee', snippet: '<marquee> element', selector: 'marquee' });
   }
   const marqueeKeyframes = collectMarqueeKeyframes(content);
   if (marqueeKeyframes.size === 0) return findings;
@@ -1082,7 +1132,7 @@ function scanCssTextForMarquee(content, markup = content) {
       const key = `${selector} ${name}`;
       if (seen.has(key)) continue;
       seen.add(key);
-      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"` });
+      findings.push({ id: 'marquee', snippet: `${selector} — infinite horizontal loop animation "${name}"`, selector });
     }
   }
   return findings;
@@ -1453,8 +1503,10 @@ function checkHtmlPatterns(html, corpora) {
   const purpleHexRe = /#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9|6366f1|764ba2|667eea)\b/gi;
   if (purpleHexRe.test(styleText)) {
     const purpleTextRe = /(?:(?:^|;)\s*color\s*:\s*(?:.*?)(?:#(?:7c3aed|8b5cf6|a855f7|9333ea|7e22ce|6d28d9))|gradient.*?#(?:7c3aed|8b5cf6|a855f7|764ba2|667eea))/gi;
-    if (purpleTextRe.test(styleText)) {
-      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected' });
+    purpleTextRe.lastIndex = 0;
+    const purpleMatch = purpleTextRe.exec(styleText);
+    if (purpleMatch) {
+      findings.push({ id: 'ai-color-palette', snippet: 'Purple/violet accent colors detected', selector: enclosingCssSelector(styleText, purpleMatch.index + 1) || undefined });
     }
   }
 
@@ -1465,7 +1517,7 @@ function checkHtmlPatterns(html, corpora) {
     const start = Math.max(0, gm.index - 200);
     const context = styleText.substring(start, gm.index + gm[0].length + 200);
     if (/gradient/i.test(context)) {
-      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient' });
+      findings.push({ id: 'gradient-text', snippet: 'background-clip: text + gradient', selector: enclosingCssSelector(styleText, gm.index) || undefined });
       break;
     }
   }
@@ -1531,7 +1583,7 @@ function checkHtmlPatterns(html, corpora) {
     const animationToken = bounceMatch[1]
       .split(/[,\s]+/)
       .find((part) => /bounce|elastic|wobble|jiggle|spring/i.test(part));
-    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}` });
+    findings.push({ id: 'bounce-easing', snippet: `animation: ${animationToken || bounceMatch[1].trim()}`, selector: enclosingCssSelector(styleText, bounceMatch.index) || undefined });
   }
 
   // Overshoot cubic-bezier
@@ -1540,7 +1592,7 @@ function checkHtmlPatterns(html, corpora) {
   while ((bm = bezierRe.exec(styleText)) !== null) {
     const y1 = parseFloat(bm[2]), y2 = parseFloat(bm[4]);
     if (y1 < -0.1 || y1 > 1.1 || y2 < -0.1 || y2 > 1.1) {
-      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})` });
+      findings.push({ id: 'bounce-easing', snippet: `cubic-bezier(${bm[1]}, ${bm[2]}, ${bm[3]}, ${bm[4]})`, selector: enclosingCssSelector(styleText, bm.index) || undefined });
       break;
     }
   }
@@ -1573,18 +1625,21 @@ function checkHtmlPatterns(html, corpora) {
 
   const glowHits = scanCssTextForGlow(styleText);
   if (glowHits.length > 0) {
-    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet });
+    findings.push({ id: 'dark-glow', snippet: glowHits[0].snippet, selector: enclosingCssSelector(styleText, glowHits[0].index) || undefined });
   }
 
   // Radial-gradient background halo (gradient-drawn sibling of dark-glow)
   const haloHits = scanCssTextForRadialHalo(styleText);
   if (haloHits.length > 0) {
-    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet });
+    findings.push({ id: 'radial-halo', snippet: haloHits[0].snippet, selector: enclosingCssSelector(styleText, haloHits[0].index) || undefined });
   }
 
   // --- Generated-UI tells: repeating-gradient stripes ---
-  if (/repeating-(?:linear|radial|conic)-gradient\s*\(/i.test(styleText)) {
-    findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes' });
+  {
+    const stripesMatch = /repeating-(?:linear|radial|conic)-gradient\s*\(/i.exec(styleText);
+    if (stripesMatch) {
+      findings.push({ id: 'repeating-stripes-gradient', snippet: 'repeating-gradient decorative stripes', selector: enclosingCssSelector(styleText, stripesMatch.index) || undefined });
+    }
   }
 
   // --- Generated-UI tells: two-axis grid-line background ---
@@ -1602,7 +1657,7 @@ function checkHtmlPatterns(html, corpora) {
   // whole gradient layers.
   const gridHits = scanCssTextForGridBackground(styleText);
   if (gridHits.length > 0) {
-    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet });
+    findings.push({ id: 'codex-grid-background', snippet: gridHits[0].snippet, selector: enclosingCssSelector(styleText, gridHits[0].index) || undefined });
   }
 
   // --- Generated-copy tells: "X theater" framing copy ---
@@ -1622,8 +1677,11 @@ function checkHtmlPatterns(html, corpora) {
   // hover:rotate / hover:translate utility on an <img>. Each distinct
   // mechanism is its own finding.
   const imgHoverCss = /\bimg\b[^,{}]*:hover\b[^{}]*\{[^}]*\btransform\s*:\s*(?:scale|rotate|translate|matrix|skew)/i;
-  if (imgHoverCss.test(styleText)) {
-    findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule' });
+  {
+    const imgHoverMatch = imgHoverCss.exec(styleText);
+    if (imgHoverMatch) {
+      findings.push({ id: 'image-hover-transform', snippet: 'img:hover { transform } rule', selector: enclosingCssSelector(styleText, imgHoverMatch.index + imgHoverMatch[0].indexOf('{') + 1) || undefined });
+    }
   }
   const imgTagRe = /<img\b[^>]*\bclass\s*=\s*"([^"]*)"/gi;
   let im;
@@ -1670,6 +1728,33 @@ function readOwnBackgroundColor(el, computedStyle) {
   return bg;
 }
 
+// One element's background-color as the cascade walk sees it: computed style
+// first (with the modern-color fallback), then, in static mode only,
+// custom-prop resolution and the inline-shorthand peek. Shared by
+// resolveBackground and resolveGradientStops so both walks read the same
+// surfaces.
+function readCascadeBackgroundColor(current, style, customPropMap) {
+  let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
+  if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
+    // The static engine can return literal "var(--X)" / "oklch(...)" strings.
+    // Resolve through customPropMap so Tailwind v4 color tokens become RGB.
+    if (customPropMap) {
+      bg = parseColorResolved(style.backgroundColor, customPropMap);
+    }
+    if (!bg || bg.a < 0.1) {
+      // Inline-style fallback for colors the static cascade did not surface
+      // on backgroundColor.
+      const rawStyle = current.getAttribute?.('style') || '';
+      const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
+      const inlineBg = bgMatch ? bgMatch[1].trim() : '';
+      if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
+        bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
+      }
+    }
+  }
+  return bg;
+}
+
 function resolveBackground(el, win, customPropMap) {
   let current = el;
   // Translucent layers (0.1 < a < 1) found on the way down to an opaque
@@ -1698,24 +1783,7 @@ function resolveBackground(el, win, customPropMap) {
     // body backgrounds.
     // Real browsers serialize wide-gamut computed values as oklab()/oklch()
     // (e.g. any color-mix() result), which plain parseRgb misses.
-    let bg = parseRgb(style.backgroundColor) || parseAnyColor(style.backgroundColor);
-    if (!DETECTOR_IS_BROWSER && (!bg || bg.a < 0.1)) {
-      // jsdom returns literal "var(--X)" / "oklch(...)" strings. Resolve
-      // through customPropMap so Tailwind v4 color tokens become RGB.
-      if (customPropMap) {
-        bg = parseColorResolved(style.backgroundColor, customPropMap);
-      }
-      if (!bg || bg.a < 0.1) {
-        // Inline-style fallback. jsdom doesn't decompose background
-        // shorthand, so colors set via inline style are otherwise invisible.
-        const rawStyle = current.getAttribute?.('style') || '';
-        const bgMatch = rawStyle.match(/background(?:-color)?\s*:\s*([^;]+)/i);
-        const inlineBg = bgMatch ? bgMatch[1].trim() : '';
-        if (inlineBg && !/gradient/i.test(inlineBg) && !/url\s*\(/i.test(inlineBg)) {
-          bg = parseColorResolved(inlineBg, customPropMap) || parseAnyColor(inlineBg);
-        }
-      }
-    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
 
     if (bg && bg.a > 0.1) {
       if (bg.a >= 0.99) return flatten(bg);
@@ -1733,40 +1801,95 @@ function resolveBackground(el, win, customPropMap) {
     //   • on other elements: bail to null and let the caller fall back
     //     to gradient stops (gradient buttons / hero sections are real
     //     bgs worth checking against).
-    if (hasGradientOrUrl) {
-      if (current.tagName === 'BODY' || current.tagName === 'HTML') {
-        return flatten({ r: 255, g: 255, b: 255, a: 1 });
-      }
-      return null;
-    }
+    // A gradient or image with no solid color under it, at any level
+    // including body/html, means the visible ground is that layer itself.
+    // Return null so the caller measures against the actual gradient stops,
+    // or skips when nothing is parseable — skipping beats a wrong ratio.
+    //
+    // Body/html used to assume white here, a guard written for jsdom, which
+    // never decomposed the `background:` shorthand and so could not see the
+    // solid paper color a texture gradient usually sits on. It turned every
+    // light-on-dark page into a wall of low-contrast false positives (a dark
+    // oklch body gradient produced ~120 "on #ffffff" findings on one site).
+    // Both engines can see shorthand solids now — the browser natively, the
+    // static cascade via expandStaticDeclaration + var() resolution — so a
+    // missing solid is real, and the old failure case cannot recur: opaque
+    // stops fully cover any hidden solid (they ARE the ground), alpha stops
+    // composite over the resolved base or the white canvas default, and
+    // unresolvable stops drop rather than guess.
+    if (hasGradientOrUrl) return null;
     current = current.parentElement;
   }
   return flatten({ r: 255, g: 255, b: 255, a: 1 });
 }
 
+// parseGradientColors (shared) reads only the legacy serializations: rgb()
+// and hex stops. Browsers keep modern-space stops in computed backgroundImage
+// exactly as authored — `linear-gradient(oklch(7% 0.006 95), …)` stays oklch —
+// which is what every token-driven page produces. Route those through
+// parseAnyColor so a gradient ground is measurable rather than invisible.
+function parseGradientColorsModern(bgImage) {
+  if (!bgImage || !/gradient/i.test(bgImage)) return [];
+  const colors = parseGradientColors(bgImage);
+  for (const m of bgImage.matchAll(/(?:oklch|oklab|hsla?|hwb)\(\s*[^()]*\)/gi)) {
+    const c = parseAnyColor(m[0]);
+    if (c) colors.push(c);
+  }
+  return colors;
+}
+
 // Walk parents looking for a gradient background and return its color stops.
 // Used as a fallback when resolveBackground() returns null because the
 // effective background is a gradient (no single solid color to compare against).
+// Translucent solid layers found between the element and the gradient (frosted
+// panels, glass washes) are composited over every stop, the same way
+// resolveBackground flattens them over a solid base — raw stops alone would
+// false-flag dark text on a light frosted wash over a dark gradient, and miss
+// the inverse.
 function resolveGradientStops(el, win, customPropMap) {
   let current = el;
+  const overlays = [];
   while (current && current.nodeType === 1) {
     const style = DETECTOR_IS_BROWSER ? getComputedStyle(current) : win.getComputedStyle(current);
     const bgImage = style.backgroundImage || '';
+    // A url() layer anywhere in the stack — alone, or alongside a gradient in
+    // the same declaration (a translucent wash over a texture photo) — paints
+    // pixels the analytic walk cannot know. Measuring the gradient stops over
+    // the wrong base flagged dark ink sitting on a bright gold-leaf image at
+    // 2.6:1; skipping beats a wrong ratio, and the screenshot subsystem owns
+    // image-backed text.
+    if (bgImage && bgImage !== 'none' && /url\s*\(/i.test(bgImage)) return null;
     let stops = null;
     if (bgImage && bgImage !== 'none' && /gradient/i.test(bgImage)) {
-      const parsed = parseGradientColors(bgImage);
+      const parsed = parseGradientColorsModern(bgImage);
       if (parsed.length > 0) stops = parsed;
     }
     if (!stops && !DETECTOR_IS_BROWSER) {
-      // jsdom doesn't decompose `background:` shorthand — peek at the raw inline style
+      // Static mode: peek at the raw inline style for gradients the cascade did not surface
       const rawStyle = current.getAttribute?.('style') || '';
       const bgMatch = rawStyle.match(/background(?:-image)?\s*:\s*([^;]+)/i);
       if (bgMatch && /gradient/i.test(bgMatch[1])) {
-        const parsed = parseGradientColors(bgMatch[1]);
+        const parsed = parseGradientColorsModern(bgMatch[1]);
         if (parsed.length > 0) stops = parsed;
       }
     }
-    if (stops) return compositeGradientStops(stops, current, win, customPropMap);
+    if (stops) {
+      const composited = compositeGradientStops(stops, current, win, customPropMap);
+      if (!composited || overlays.length === 0) return composited;
+      return composited.map(stop => {
+        let acc = stop;
+        for (let i = overlays.length - 1; i >= 0; i--) acc = compositeColorOver(overlays[i], acc);
+        return acc;
+      });
+    }
+    const bg = readCascadeBackgroundColor(current, style, customPropMap);
+    if (bg && bg.a > 0.1) {
+      // An opaque surface above the gradient means the gradient never shows
+      // through here; resolveBackground would have returned it, so reaching
+      // this is defensive — bail rather than measure the wrong layer.
+      if (bg.a >= 0.99) return null;
+      overlays.push(bg);
+    }
     current = current.parentElement;
   }
   return null;
@@ -1986,6 +2109,10 @@ function checkElementColorsDOM(el) {
   const rect = el.getBoundingClientRect();
   if (rect.width < 10 || rect.height < 10) return [];
   const style = getComputedStyle(el);
+  // Invisible at rest: hidden scene variants (opacity-0 carousels, swap
+  // decks) are not user-visible, and measuring their inherited colors against
+  // whatever surface happens to sit behind the stack is noise, not audit.
+  if (style.visibility === 'hidden' || effectiveOpacityDOM(el) <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
   let effectiveBg = resolveBackground(el);
@@ -2798,11 +2925,13 @@ function checkElementGlowDOM(el) {
   // If resolveBackground returns null (gradient), try to infer from the gradient colors
   let parentBg = el.parentElement ? resolveBackground(el.parentElement) : resolveBackground(el);
   if (!parentBg) {
-    // Gradient background — sample its colors to determine if it's dark
+    // Gradient background — sample its colors to determine if it's dark.
+    // Modern-syntax parsing matters here: body-level gradients now reach this
+    // fallback in browser mode, and their stops usually serialize as oklch.
     let cur = el.parentElement;
     while (cur && cur.nodeType === 1) {
       const bgImage = getComputedStyle(cur).backgroundImage || '';
-      const gradColors = parseGradientColors(bgImage);
+      const gradColors = parseGradientColorsModern(bgImage);
       if (gradColors.length > 0) {
         // Average the gradient colors
         const avg = { r: 0, g: 0, b: 0 };
@@ -3644,6 +3773,14 @@ function checkElementBorders(tag, style, overrides, resolvedRadius, el = null) {
 }
 
 function checkElementColors(el, style, tag, window, customPropMap, hasAnchorInheritRule) {
+  // Invisible at rest, static twin of the browser walk's skip: opacity does
+  // not inherit, so walk ancestors multiplying declared opacity down.
+  if (style.visibility === 'hidden') return [];
+  let effOpacity = 1;
+  for (let cur = el; cur && cur.nodeType === 1 && effOpacity > 0.02; cur = cur.parentElement) {
+    effOpacity *= parseFloat(window.getComputedStyle(cur).opacity || '1');
+  }
+  if (effOpacity <= 0.02) return [];
   const directText = [...el.childNodes].filter(n => n.nodeType === 3).map(n => n.textContent).join('');
   const hasDirectText = directText.trim().length > 0;
 
@@ -4802,6 +4939,11 @@ function isRenderedForBrowserRule(el) {
 function checkElementTextOverflowDOM(el) {
   const tag = el.tagName.toLowerCase();
   if (TEXT_OVERFLOW_SKIP_TAGS.has(tag)) return [];
+  // scrollWidth/clientWidth are CSS box-model metrics; on SVG content Chrome
+  // returns arbitrary non-zero values for both (a <text> reported 78/48 while
+  // its rendered length sat comfortably inside its box), so the delta is
+  // noise, not overflow. SVG clips to its own viewport anyway.
+  if (el.namespaceURI === 'http://www.w3.org/2000/svg') return [];
   if (!isRenderedForBrowserRule(el)) return [];
   // Only the element that actually owns overflowing text — not its ancestors,
   // which inherit a wider scrollWidth from the spilling descendant.
@@ -5186,6 +5328,22 @@ function isPaintedForOcclusion(el) {
 // path is pure geometry and runs anywhere on the page.
 const OCCLUSION_TEXT_SKIP_TAGS = new Set(['script', 'style', 'noscript', 'template', 'title']);
 
+// An element whose effective opacity multiplies out to ~0 paints nothing at
+// rest: it is not user-visible, so visual findings on it (contrast, occlusion)
+// measure a state nobody sees. Browser-only — the walk needs live computed
+// styles. Cycling scenes that fade such elements in later are the screenshot
+// subsystem's territory, not the analytic walk's.
+function effectiveOpacityDOM(el) {
+  let o = 1;
+  // Walk all the way through body and html: `body { opacity: 0 }` page-fade
+  // wrappers hide every descendant just as thoroughly as a local wrapper.
+  for (let cur = el; cur && cur.nodeType === 1; cur = cur.parentElement) {
+    o *= parseFloat(getComputedStyle(cur).opacity || '1');
+    if (o <= 0.02) return 0;
+  }
+  return o;
+}
+
 function checkTextOcclusionDOM() {
   const findings = [];
   const seenVictims = new Set();
@@ -5213,6 +5371,11 @@ function checkTextOcclusionDOM() {
     }
     return false;
   };
+  // The classic occluder shape this rules out is an opacity-0 interaction
+  // layer — a range scrubber stretched over a before/after comparison — which
+  // elementFromPoint still returns and whose UA background-color otherwise
+  // reads as an opaque box.
+  const effectiveOpacity = effectiveOpacityDOM;
 
   // Collect renderable text owners in / near the first viewport for the
   // elementFromPoint probe. SVG <text> counts too.
@@ -5225,6 +5388,7 @@ function checkTextOcclusionDOM() {
     const text = inSvg ? (el.textContent || '').trim() : elementDirectText(el);
     if (text.length < 2) continue;
     if (!isPaintedForOcclusion(el)) continue;
+    if (effectiveOpacity(el) <= 0.02) continue;
     let rect; try { rect = el.getBoundingClientRect(); } catch { continue; }
     if (rect.width < 6 || rect.height < 6) continue;
     // Viewport-bound probe: keep text whose box overlaps the live viewport.
@@ -5258,6 +5422,7 @@ function checkTextOcclusionDOM() {
         if (top === el || el.contains(top) || top.contains(el)) continue;
         const topCs = getComputedStyle(top);
         if (isFloated(topCs) || isMarqueeish(top, topCs) || isPinnedOverlay(top)) continue;
+        if (effectiveOpacity(top) <= 0.02) continue;
         const topTag = top.tagName.toLowerCase();
         // Text sitting under a raw image/video is contrast territory (deduped
         // against the pixel low-contrast rule); leave those alone here.
@@ -5468,6 +5633,7 @@ export {
   CSS_NAMED_COLORS,
   checkBorders,
   isEmojiOnlyText,
+  scopedIgnoreActive,
   checkColors,
   checkHoverContrast,
   checkElementHoverContrast,