gemini-3-pro-image is Google's strongest image model; IMAGE_GEN_MODEL still overrides for the cheaper flash tier.
AI-assisted commit.
Co-authored-by: Cursor <cursoragent@cursor.com>
Ban hex codes from image prompts to stop Nano Banana swatch strips, require crisp deep focus and set-dressed surfaces inside close product-world scenes, and add wordless and real-estate inspection retries. The Gemini wrapper now converts JPEG bytes to PNG and infers AQ.-prefixed keys.
AI-assisted commit.
Co-authored-by: Cursor <cursoragent@cursor.com>
Ship image-gen.mjs with bfl and gemini backends plus custom delegation, document the keyless flow in image-api.md, and tighten seed Step 4 so missing keys ask once and opt-out falls back to text-only seeding. Compile now removes in-folder hero intermediates so each cue keeps one PNG.
AI-assisted commit.
Co-authored-by: Cursor <cursoragent@cursor.com>
Each specialist now generates one hero image instead of a second chroma-key sheet, and the compile step replaces crop with a simpler cues.json shape.
AI-assisted commit.
Co-authored-by: Cursor <cursoragent@cursor.com>
Drop the two-pass matte pipeline in favor of one chroma-key sheet per
cue: each specialist picks the farthest-hue key, stages larger artifacts
with shadowless crisp edges, reports CHROMA in its reply, and crop records
the hex in cues.json for browser keying downstream.
Co-authored-by: Cursor <cursoragent@cursor.com>
Artifact crops now ship with real alpha: each sheet is generated twice
(cream, then black via image edit) and fused by a new difference-matting
command, sidestepping models' inability to emit alpha and the chroma
key's spill. Crop and matte reject non-square inputs so squareness must
come from the generation call's 1:1 parameter, and the hero prompt now
demands bright, generous light so palettes stay sampleable.
Co-authored-by: Cursor <cursoragent@cursor.com>
A live run shipped one hero under three slugs: parallel subagents all
wrote the generation tool's output as a generic hero.png in the shared
project assets folder, then copied and referenced each other's files.
Spawn tasks now require slug-based output filenames, using only the
tool-reported path, a verify-your-own-files check before reporting, and
the parent hashes all heroes before cropping and re-spawns duplicates.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Live cues rendered the good colors as slivers: the hero prompt asked for
an atmospheric scene with a 60% neutral ground and a small accent. The
skeleton now frames tight on the subject and stages each color as a
large unbroken field (neutral about half, primary a continuous third,
tertiary small but legible), with the rationale in the fill rules.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Investigation of a live run showed five of six palettes converging on the
brief's deep-teal anchor: territories were scenic but hue-silent, and the
anchor reached every spawn unscoped. Territories now must close on a hue
ground (no two in one family), the anchor is owned by exactly one
territory with everyone else told to leave it, spawn tasks may not be
compressed, and each subagent critiques its palette and concept against
the rules before generating.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Name the common spawn tools, require all six spawns in a single message,
and call out spawn-await-spawn as the serial-loop failure it is. The
no-subagent fallback now runs all six personas instead of four, so the
user gets six cues on every path.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Drop the sequential palette chain: the parent carves six visual,
emotional, or strategic territories from the brief, then all six
personas spawn in one parallel batch, each composing inside its own
territory and generating both images in the same task.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Personas now compose sequentially, each shown the color territories the
earlier personas claimed, replacing the similarity script and peer
critique as the uniqueness mechanism. Generation moves to a parallel wave
of dedicated image-prompt specialists that stage the finished palettes.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Drop the six-way critique wave from the visual-cues studio: after the
compose wave, the OKLab similarity script loops with revision waves for
flagged personas only (capped at three) until every verdict is clear,
then the generation wave stages the settled palettes.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Drop the parallel "round" vocabulary from the visual-cues studio so one
counting system covers the three spawn waves; the subagent template calls
the palette a first draft instead.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Replace the single-mind palette composition in the visual-cues pipeline
with a three-wave studio: six persona specialists compose in parallel, a
new OKLab similarity command plus peer critiques build each persona's
scorecard, then each revises and generates its hero and artifact sheet.
AI-assisted (Cursor agent), directed by abdulwahabone.
Co-authored-by: Cursor <cursoragent@cursor.com>
Document seed mode's palette pick used to ask the user to imagine colors
from words. Add a visual-cues pipeline: compose four-hex palettes from the
brief (mood, references, PRODUCT.md), generate a hero scene plus a cream
artifact sheet per concept, crop the sheet into isolated artifacts, and
write cues.json so the user can pick a direction by eye.
Co-authored-by: Cursor <cursoragent@cursor.com>
A stored IMAGE_GEN_API_KEY in .impeccable/.env now counts as image generation being available for shape probes and the craft codex flow. The key ask itself stays exclusive to document seed mode; shape and craft still never prompt for APIs or tooling.
Co-authored-by: Cursor <cursoragent@cursor.com>
Seed questions now avoid skill vocabulary (mirroring init.md's magazine-editor line), and a new step offers 2-4 generated palette cue images: native image_gen when the harness has it, otherwise an opt-in IMAGE_GEN_API_KEY in gitignored .impeccable/.env.
Co-authored-by: Cursor <cursoragent@cursor.com>
Ask for logos, reference images, and moodboards before the five-question interview so the first DESIGN.md seed can ground taste selections in real visual context.
Co-authored-by: Cursor <cursoragent@cursor.com>
Prefer committed visual goldens when the app cannot run and make assigned-system translation explicit when a pinned register conflicts with literal materials.\n\nAI-assisted: prepared with Codex under @pbakaus direction.
Preserve critique trend history while closing the exact processed snapshot. Track target identity and content freshness for local files and URLs, isolate colliding streams, and prevent legacy snapshots from resurfacing after a modern close.
Validated with 50 focused tests, a clean 17-provider build, 834 core tests, passing GitHub CI and Cursor Bugbot, Greptile 5/5, and zero unresolved review threads.
AI-assisted maintainer repair: implemented and reviewed with Codex under @pbakaus direction.
Store direction contracts in development-only surface briefs and guard against browser-delivered leakage with regression coverage.
AI-assisted merge: reviewed and executed by Codex under maintainer direction.
Paul's decision on the tenth sweep's design question: hard vetoes (missing region, contradicted plate or text, SVG illustration, clipped plate, invented ink) stay unconditional; at overall >= HERO_MIN the numeric readings (ink colour, letter-spacing, line pitch, strip heights, box positions) print as advisories with the pass and belong to the polish pass. Every sweep-10 sample closes its hero under this condition, which settles 07 without another paid round.
Ninth-sweep defects: codeDrawn / container / bleed now persist into spec.json with WARN lines (an overridden refusal used to vanish from the record); font-match probes os.tmpdir() and points TMPDIR at .impeccable/tmp when the sandbox /tmp is unwritable (every ninth-sweep rank silently fell back to the catalog).
AI-assisted (Claude Code).
A reference, not the page: --r-<id>-x/y/w/h in % of the comp (plus cap height, font-size, family, weight where measured) to bind to any markup, and hero-reference.html with every region at its box and every plate placed with object-fit: contain, as a check on positions. Attacks the most common execution failure of weaker builders (badly positioned, overflowing, pushed below the fold) without dictating structure to strong ones; overlapping boxes are overlapping boxes and the gate reads pixels regardless.
AI-assisted (Claude Code).
From the human review's most repeated pin ('terrible svg instead of asset', on every model) and from sessions that wrote 'ship' with the hero open. Icons, arrows, chevrons, and runtime data charts stay code; diagrams, notation, and leader lines are plates.
AI-assisted (Claude Code).
From the final review batch: 'bad asset crop (crops are never allowed)' twice, 'letter spacing way too wide'. A crop resampled to the region scores 99.8% structure against the raw region; a produced plate scores 30-60.
AI-assisted (Claude Code).
A session's own note said its hero sat at 67 because the 10x10 grid boxes straddled two elements each, and it was right: every downstream measurement (cap height, line count, structure) inherited the slop. Text and control regions now snap to the dominant connected ink inside the span (page-ground threshold, dilated cells, masses touching the span's sides lose to inside masses), keep the span on the record for coverage, and can opt out with snap: false.
AI-assisted (Claude Code).
From the second review batch: the best build of the fifth sweep passed the hero at 87% with the cover arch cut flat on the left (object-fit: cover on a box narrower than the shape), and legends, badges, and extra controls one or two cells wide slipped under the invented-ink floor. comp-spec measures the artwork's contiguous contact with each box edge against the page ground; the spec gate refuses such a box unless bleed is set. Human pass line landed at comp-diff 72-73; HERO_MIN stays 0.72.
AI-assisted (Claude Code).
From the first human review of sweep-3 builds (pins on 12 samples): fonts at the wrong size, weight, colour, or place; nav bars too tall; kickers and dividers the comp does not have; a footer strip pushed off the frame that read as drift on ground colour alone; a drawing filed as chrome with no note to catch it.
- lib/hero-checks.mjs: textRegionCheck (cap height, line count, ink density, ink colour, first-line offset vs the comp crop, measured fresh), chromeStripCheck (first rule row), inventedInk (build energy over a calm comp cell and neighbourhood). Wired into gateHero as reasons; invented ink vetoes at 4% of cells.
- comp-diff verdictFor: detailRaw < 0.15 is missing whatever the palette.
- comp-spec: every region carries a note.
- font-fingerprint: the tall-line filter takes its median over lines with real mass, so two display lines above a small line are not 'tall'.
AI-assisted (Claude Code).
A session named seven regions for a page with three plates, a table, a note, callouts and a spine, so the hero gate could name nothing and the score sat at 70. A code region is one element; a column is a container of several.
AI-assisted (Claude Code).
font-match / catalog index
- Index schema 2 adds a third render, 48c (48px cap, ALL CAPS text). Caps headline crops have no x-height band and ranked against mixed-case renders as barcode faces; they now route to the caps render.
- Non-text families (barcodes, redacted, flow, dingbats, effect faces) are excluded from candidates.
- The distance adds a gross width and weight gap (log ratio of advance and ink density) so a face 50% wider or 35% lighter cannot rank first on run-length detail alone; the index stores those readings.
- Multi-line all-caps crops vote on x-height across lines: one line's crossbars no longer give the crop a spurious x band.
- With no browser, --rank records the catalog's nearest face (source catalog, size estimated) so the spec gate can close; the NEXT line and new-work.md say not to install a browser or hand-write a choice.
- font-match stamps the choice it writes; the spec gate refuses a chosen face it did not write (sessions typed Arial Narrow into spec.json to pass).
- IMPECCABLE_NODE_MODULES lets a harness lend a playwright.
comp-spec
- A region note that describes painted material (diagram, drawing, photo, texture...) under a code kind is refused at the spec unless codeDrawn is set: the exploded carburetor filed as chrome is a plate about to be redrawn in SVG.
build-phase
- The control ink-box veto applies only when the comp's ink is a discrete element and the build's box is too; a full-width bar told one session six times that 1376x87 was 1382x102 with no edit able to move it.
AI-assisted (Claude Code).
lib/font-fingerprint.mjs replaces the three-number fingerprint with
size-invariant shape features (x-height ratio, stroke contrast, stem width,
run-length quantiles, roundness, serif signal, width spread) and a
noise-normalized distance; family recall on a held-out self-test rose from
13% to 72% top-5. data/font-index.json carries the whole Google Fonts
catalog (3,092 faces at two cap sizes, 707 KB); font-match --rank fingerprints
the comp crop, takes the 25 nearest faces from the index (plus the model's
own names), renders them at the comp's cap height, ranks by the same
distance, and prints a proof sheet and the CSS to use. scripts/build-font-
index.mjs rebuilds the index at release time.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
The first paid confirmation sweep showed the failure: models rendered the
three comps first and ran build-phase.mjs start after, so a session cut at
the composition pick carried no state.json and the resumed model followed
the conversation ('translate the comp into HTML now') instead of the
phases. Decision comps (.impeccable/mocks/decision/) are unaffected;
--force-mock overrides.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
font-match.mjs fingerprints a text region's lettering in the comp (cap
height, glyph advance per cap = width class, ink fraction = weight class,
tracking), renders candidate faces at that cap height in a headless
browser (yours plus a shortlist for the width class), and ranks them by
distance with a proof sheet; the spec gate refuses to close until the lead
text region is measured and ranked. generate-image --plate keys ink-on-
ground plates to alpha (chroma) so the drawing sits on the page's own
ground; the plates gate scores keyed plates composited over the region's
ground. comp-diff reports each region's ink box; the hero gate names a
control whose box height or width differs from the comp. The hero
instruction copies the comp's words verbatim; rewording is a stated
decision after the hero passes.
Driven by a human review of the r3 side-by-sides: face width and weight,
plate ground, control row height, and content substitution.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
Round-1 simulated builds passed the hero at 1536 and shipped a page whose
first viewport collapsed to one column at 1440 (comp-diff 50% on the
final capture, 82% on the hero). The responsive phase now requires
desktop.png and mobile.png and diffs desktop.png against the comp at 65%
with no region missing.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
From a forensics pass over twelve runs and two adversarial passes over
the metrics:
- comp-diff: detail is signed and penalizes invented energy; regions with
structure under 0.3 (or painted regions under 0.45 / added detail over
0.4) are contradicted whatever the mean says; palette ramp tightened;
region crops inherit the whole-image best translation so a shifted page
is not eight contradicted regions.
- hero gate: fails on any contradicted plate/image/text region (chrome and
controls keep the one-third allowance), on a capture that is not the
comp's frame, on a palette that is not the comp's, and on an organic
clip-path drawn inside a raster region's box.
- plates gate: scored against the comp crop with overlapping text/chrome
painted out (comp-spec plateReference; generate-image uses it too), with
a structure floor and an added-detail veto; the real plate passes, noise,
mirrors, mosaics, and other regions do not.
- comp-spec refuses a regions file that leaves comp ink unnamed.
- The direction-choice ping is folded into build-phase.mjs start
--direction --kind; the roll writes .impeccable/build/pending.json and
context.mjs / detect.mjs report COMP_ROUND_OPEN until the hero gate
passes. A code-led config makes start print the contract step and stop.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
concept-seed's choice ping now prints the next mandatory step from the
recorded build path (comp-led: build-phase.mjs start --direction <key>),
because every run that skipped the comp round did so right after that
ping. build-phase gains a comps phase ahead of spec (three sidecar'd comps
under .impeccable/mocks/, one approved) and records the approved comp on
close. The hero gate lists the worst region crops first with the fix class
per verdict, and refuses a third value-only attempt on the same stuck
region. Hero instruction is plates first, then the semantic layer. The
finish reviewer treats a comp-led build with no closed comps phase as a
material finding.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
The first live run produced a faithful carburetor plate, then drew the
region in SVG and left the plate on disk. Before diffing, the hero gate
now walks the artifact (or a bounded source tree) for every plate's file
name or a data URI named for it.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
The first live run forced past the plates gate with 'single-file HTML
delivery requires embedded CSS/SVG'. That is not a reason the comp's
authority moves for; the script now refuses a --force whose reason does
not quote the user, and new-work.md says a single-file deliverable inlines
the plate as a data URI. Texture plates are judged on palette and grain,
not size or structure, since they tile.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
new-work.md section 6 becomes the spec / plates / hero / sections / motion /
responsive phase list, each closed by build-phase.mjs advance; the
reproduction and hero-checkpoint prose that asked the model to compare
from memory is gone. visualize.md's inventory, medium gate, and produce
sections collapse into the spec and plate mechanism. The finish reviewer
takes the state file and comp-diff reports as inputs and starts its
fidelity matrix from the measured verdicts. docs/COMP-FIDELITY.md records
the design.
AI-assisted (Claude).
Co-Authored-By: Claude <noreply@anthropic.com>
Grok was classified as GitHub Copilot, so the design hook skipped every
edit with no-file-path and never ran Stop. Normalize toolInput/sessionId
and treat Stop additionalContext as the Grok product.
Prepared with AI assistance.
Co-authored-by: Cursor <cursoragent@cursor.com>
Claude Code now folds multi-edit behavior into Edit, so keep generated and repaired hook manifests aligned with the current Edit and Write tools. Grok keeps its compatibility matcher unchanged.
AI assistance was used to implement and validate this change.
Abdul WahabandCursor
Paul BakausandGitHub
Abdul WahabandGitHub
MorganandGitHub
