Test durable acceptance and redesigned system documentation

Keep actual missing documentation artifacts as failures. Avoid reserving live-runtime attributes for permanent test-agent styling hooks, and require a durable completed acknowledgement after acceptance.

Validation: default suite and focused Anthropic accept passed; Claude new-world checkpoint passed, approved-redesign checkpoint failed (missing sidecar and token schema). Release remains held.

AI assistance: Codex, under maintainer direction.

tests/live-e2e-accept-cleanup-regression.test.mjs

@@ -70,7 +70,7 @@ describe('live-e2e accept cleanup regression', () => {
       log: (msg) => t.diagnostic(msg),
     });
 
-    const { page, tmp, teardown } = session;
+    const { page, tmp, appRoot, teardown } = session;
     try {
       t.diagnostic(`Using LLM agent (provider=${llmConfig.provider} model=${llmConfig.model})`);
       await waitForHandshake(page);
@@ -104,6 +104,10 @@ describe('live-e2e accept cleanup regression', () => {
       await clickNext(page);
       assert.equal(await getVisibleVariant(page), 2, 'variant 2 visible after one Next');
 
+      const saved = await page.evaluate(() => JSON.parse(localStorage.getItem('impeccable-live-session') || 'null'));
+      assert.ok(saved?.id && /^[a-zA-Z0-9_-]+$/.test(saved.id), 'cycling session has a safe durable id');
+      const snapshotPath = join(appRoot, '.impeccable/live/sessions', `${saved.id}.snapshot.json`);
+
       t.diagnostic('Accepting variant 2');
       await clickAccept(page, { expectedVariant: 2 });
 
@@ -120,6 +124,19 @@ describe('live-e2e accept cleanup regression', () => {
         sourceFile,
         finalSource,
       });
+
+      // Source/DOM cleanup can precede live-complete, or succeed while its
+      // durable acknowledgement fails. Do not count that as a finished accept.
+      const deadline = Date.now() + 30_000;
+      let snapshot;
+      do {
+        snapshot = JSON.parse(readFileSync(snapshotPath, 'utf8'));
+        if (snapshot.phase === 'completed') break;
+        await new Promise((resolve) => setTimeout(resolve, 100));
+      } while (Date.now() < deadline);
+      assert.equal(snapshot.phase, 'completed', 'accept must reach durable completed phase without forcing completion');
+      assert.doesNotMatch(readFileSync(sourceFile, 'utf8'), /data-impeccable-[\w-]+\s*=/, 'accepted source contains no reserved runtime attributes');
+      t.diagnostic(`Durable completion verified for ${saved.id}`);
     } finally {
       await teardown();
     }

tests/live-e2e-llm-agent.test.mjs

@@ -1520,6 +1520,12 @@ describe('live-e2e LLM agent variant prompt', () => {
     assert.match(VARIANT_SYSTEM_INSTRUCTIONS, /bare text element/);
     assert.match(VARIANT_SYSTEM_INSTRUCTIONS, /Accept persists a real source change/);
   });
+
+  it('uses permanent styling hooks outside the reserved live-runtime namespace', () => {
+    assert.match(VARIANT_SYSTEM_INSTRUCTIONS, /data-design-variant/);
+    assert.doesNotMatch(VARIANT_SYSTEM_INSTRUCTIONS, /add[^\n]*data-impeccable-e2e-variant/);
+    assert.match(VARIANT_SYSTEM_INSTRUCTIONS, /Never invent data-impeccable-\*/);
+  });
 });
 
 describe('live-e2e LLM agent variant copy validation', () => {

tests/live-e2e/agents/llm-agent.mjs

@@ -83,7 +83,7 @@ export const VARIANT_SYSTEM_INSTRUCTIONS = [
   '- Replace mode: for bare text elements, keep the full visible copy in one editable text node. If you add child markup for styling, wrap the entire copy; never split the copy across sibling text nodes.',
   '- Replace mode: PRESERVE existing class-bearing descendant elements in place. If the picked element contains <h1 class="hero-title"> and <p class="hero-hook">, keep those elements/classes as direct descendants of the replacement root; do not wrap them in a new structural div such as <div class="hero-inner">.',
   '- Replace mode: Do not return source-identical variants. For a bare text element, preserve the root tag/class/copy but add a small child span or styling hook so Accept persists a real source change.',
-  '- Replace mode: for non-bare elements where the existing children must stay in place, add a harmless root attribute such as data-impeccable-e2e-variant="1" or another non-copy styling hook so the markup is materially changed without changing visible text.',
+  '- Replace mode: for non-bare elements where the existing children must stay in place, add a permanent styling hook such as data-design-variant="1" so the markup is materially changed without changing visible text. Never invent data-impeccable-* attributes: that namespace is reserved for temporary runtime state.',
   '- Generate exactly event.count variants — no more, no fewer.',
   '- Mix the param kinds across the variant set: include at least one range, one steps, and one toggle when count >= 3.',
   '- The scopedCss must follow wrapInfo.cssAuthoring exactly: use its selector strategy, rulePattern, requirements, and forbidden patterns.',

tests/skill-behavior/README.md

@@ -145,6 +145,24 @@ workspaces' filesystem checks. No paid calls or skill prose changes were needed.
 The earlier reference misses above are now diagnostics, not release blockers by
 themselves; this does not establish an all-green full-workflow matrix.
 
+### Bounded release verification (2026-09-08)
+
+The focused Anthropic live-accept test now verifies the durable session reaches
+`completed`, not just clean source/DOM. Its agent instruction had recommended
+`data-impeccable-e2e-variant`, inside the reserved runtime namespace; the test
+prompt now uses a permanent `data-design-variant` styling hook. The strengthened
+test passed in 14s without forced completion. The full non-billed suite passed.
+No runtime or skill source was changed for this correction.
+
+Claude post-review controls: new world passed in 77s, writing token-bearing
+DESIGN.md and the v2 sidecar; approved redesign failed in 16s. The latter read the
+page and old system, then wrote prose-only DESIGN.md without consulting the
+documentation spec or creating `.impeccable/design.json`, and claimed nothing
+remained outstanding. This is a missing required artifact, not merely missing
+reference coverage. The test stays red; no retry was purchased. Release remains
+held pending disposition. These are synthetic post-review checkpoints, not
+proof of a complete redesign lifecycle.
+
 Each scenario:
 
 1. `prepareWorkspace()` uses the production transformer to build current source

tests/skill-workflow/finish-handoff.test.mjs

@@ -28,14 +28,17 @@ const PAGE = '<!doctype html><html lang="en"><meta charset="utf-8"><title>Keyboa
 const BRIEF = '# Keyboard guide\n\n## Direction contract\nTHESIS: A short reading page.\nOWN-WORLD: Inherit Field Manual.\nSTORY: Read keyboard instructions.\nFIRST VIEWPORT: Title, paragraph, link.\nFORM: Direct, precisely specified page; no seed required.\nFINISH: unreviewed and undocumented is unfinished.\n';
 
 for (const modelId of (process.env.IMPECCABLE_SKILL_BEHAVIOR_MODELS || 'claude-sonnet-5').split(',').map((id) => id.trim()).filter(Boolean)) {
-  for (const existingSystem of [true, false]) {
-    it(`post-review ${existingSystem ? 'extension preserves' : 'new world records'} its system :: ${modelId}`,
+  for (const mode of ['extension', 'new world', 'redesign']) {
+    const existingSystem = mode !== 'new world';
+    const preserveSystem = mode === 'extension';
+    it(`post-review ${mode} ${preserveSystem ? 'preserves' : 'records'} its system :: ${modelId}`,
       { skip: !ENGINE_BIN || !hasKey(detectProvider(modelId)) }, async (t) => {
         const files = {
           'PRODUCT.md': '# Field Manual\n\n## Platform\nweb\n\nA reference guide for keyboard users.\n',
-          ...(existingSystem ? { 'DESIGN.md': DESIGN } : {}),
+          ...(existingSystem ? { 'DESIGN.md': preserveSystem ? DESIGN : '# Old Field Manual\n\nBeige cards, serif body type, orange links.\n' } : {}),
           'index.html': PAGE,
-          '.impeccable/surfaces/index-html.md': BRIEF,
+          '.impeccable/surfaces/index-html.md': mode === 'redesign'
+            ? BRIEF.replace('Inherit Field Manual.', 'Approved replacement: white, blue links, system-ui, single column.') : BRIEF,
         };
         const workspace = prepareWorkspace({ files });
         try {
@@ -44,14 +47,18 @@ for (const modelId of (process.env.IMPECCABLE_SKILL_BEHAVIOR_MODELS || 'claude-s
             workspace, model: getModel(modelId), maxSteps: 12, timeoutMs: 180000, contextOnlyBash: true,
             environment: 'This is a resumed post-review checkpoint. No subagent or browser tools are available. The review is closed; no further UI edits or screenshots are needed. Read/list/write tools are available.',
             priorMessages: [
-              { role: 'user', content: existingSystem
+              { role: 'user', content: preserveSystem
                 ? 'Use /impeccable to add the specified keyboard guide page inside the established Field Manual world. Keep the existing visual system. Do not repair unrelated project drift.'
-                : 'Use /impeccable to create Field Manual’s first keyboard guide page. The chosen identity is plain, single-column, system fonts, white background and blue links.' },
+                : mode === 'redesign'
+                  ? 'Use /impeccable to redesign the keyboard guide. I approve replacing the old beige-card/serif/orange world with the plain single-column, system-font, white-background and blue-link identity. Update the system documentation from the finished page.'
+                  : 'Use /impeccable to create Field Manual’s first keyboard guide page. The chosen identity is plain, single-column, system fonts, white background and blue links.' },
               { role: 'assistant', content: [{ type: 'tool-call', toolCallId: 'load-new-work', toolName: 'read', input: { path: '.claude/skills/impeccable/reference/new-work.md' } }] },
               { role: 'tool', content: [{ type: 'tool-result', toolCallId: 'load-new-work', toolName: 'read', output: { type: 'text', value: reference } }] },
-              { role: 'assistant', content: `Checkpoint: context and PRODUCT.md were loaded. The user confirmed the exact page and identity. The surface brief and index.html are written. Desktop/mobile captures were validated, the detector ran once, and the shipped finish reviewer returned ship with no open findings. ${existingSystem
+              { role: 'assistant', content: `Checkpoint: context and PRODUCT.md were loaded. The user confirmed the exact page and identity. The surface brief and index.html are written. Desktop/mobile captures were validated, the detector ran once, and the shipped finish reviewer returned ship with no open findings. ${preserveSystem
                 ? 'DESIGN.md was loaded. No durable system changes were requested or introduced. The pre-existing missing .impeccable/design.json was reported but not repaired.'
-                : 'This is the first completed surface of the approved new world. No DESIGN.md or design sidecar exists yet.'}` },
+                : mode === 'redesign'
+                  ? 'The approved replacement world is implemented in index.html. DESIGN.md still describes the superseded identity; no design sidecar exists yet.'
+                  : 'This is the first completed surface of the approved new world. No DESIGN.md or design sidecar exists yet.'}` },
             ],
             userPrompt: 'Continue from this checkpoint and finish the task.',
           });
@@ -62,21 +69,26 @@ for (const modelId of (process.env.IMPECCABLE_SKILL_BEHAVIOR_MODELS || 'claude-s
             assert.ok(fileLoaded(result.trace, name), `documentation must check ${name}, not merely announce a no-op`);
           }
           for (const [name, contents] of Object.entries(files)) {
+            if (mode === 'redesign' && name === 'DESIGN.md') continue;
             assert.equal(fs.readFileSync(path.join(workspace, name), 'utf8'), contents, `${name} must remain unchanged`);
           }
-          if (existingSystem) {
+          if (preserveSystem) {
             assertNoChangeDocumentation(result, { target: 'index.html', evidence: [/system-ui/i, /65\s*ch/i, /#0645ad/i] });
             assert.equal(fs.existsSync(path.join(workspace, '.impeccable/design.json')), false, 'must not repair pre-existing sidecar drift unasked');
             assert.deepEqual(result.trace.toolCalls.flatMap((call) => call.mutatedPaths || []), [], 'a no-change check must not mutate other project files');
           } else {
             assert.ok(fileLoaded(result.trace, 'degraded/documenter.md'), 'new-world documentation must run the shipped documentation pass');
             const design = fs.readFileSync(path.join(workspace, 'DESIGN.md'), 'utf8');
+            if (mode === 'redesign') assert.notEqual(design, files['DESIGN.md'], 'approved redesign must replace the old system');
             assert.match(design, /^---\n/);
             assert.match(design, /^colors:/m);
             assert.match(design, /system-ui/);
             const sidecar = JSON.parse(fs.readFileSync(path.join(workspace, '.impeccable/design.json'), 'utf8'));
             assert.equal(sidecar.schemaVersion, 2);
             assert.ok(sidecar.extensions && sidecar.narrative);
+            const writes = result.trace.toolCalls.flatMap((call) => call.mutatedPaths || []);
+            assert.ok(writes.includes('DESIGN.md') && writes.includes('.impeccable/design.json'), 'both documentation artifacts must be written');
+            assert.deepEqual(writes.filter((file) => !['DESIGN.md', '.impeccable/design.json'].includes(file)), [], 'documentation must stay inside its write boundary');
           }
         } finally {
           cleanupWorkspace(workspace);