feat(spec)!: ai:chat_window is retired — refused by name at the schema door, the floating chat overlay is the AI chat entry point (#21504)

[objectstack-fleet]

Fixes #21504
Clause-②: yes (narrowing)

Retires the ai:chat_window page component type under ADR-0049 enforce-or-remove, refused by name through RETIRED_PAGE_COMPONENT_TYPES, per triage ruling 5963897014 and the user:profile precedent (#14159, landed in PR #15112). No renderer for it ever shipped: objectui leaves it unregistered on purpose, because the floating chat overlay is the supported AI chat entry point. So an authored node validated clean and then drew "Unknown component type". ai:suggestion is not touched; triage ruled it a different class.

What changed

surface	change
ui/page.zod.ts	'ai:chat_window' leaves PageComponentType and joins RETIRED_PAGE_COMPONENT_TYPES. Its message names the floating chat overlay as the supported entry point and carries no tracker number. PageComponentSchema.type already consults the map, so the open string arm cannot re-admit the name. The type's .describe() and the enum docblock name the new member.
ui/component.zod.ts	AIChatWindowProps (mode, agentId, context, aria) is deleted. The ComponentPropsMap['ai:chat_window'] row stays as retiredComponentProps('ai:chat_window'): a z.never that refuses every bag, {} included, with the same prescription.
ADR-0087	D3 semantic entry ui-ai-chat-window-retired, one STEP18_RATIONALE fragment, and the RETIRED_DEFS_BY_MAJOR[18] entry ui/AIChatWindowProps, generated by gen:migration-registry. No D2 conversion, see below.
pins	component.test.ts gets a new describe with code, path, params and the first sentence at each door: the map, the row (empty and populated bags), PageComponentSchema, PageSchema at the element path, and the enum error map. ai:suggestion is the control and parses at every door, and the open arm stays open. component-type-vocabulary.test.ts: the type stays known, leaves the typo candidates, and ai: stays reserved. The old should parse ai:chat_window with default accept pin is flipped into the new describe.
enumeration pin	component-props-unknown-members.pin.test.ts (PR #21502) loses its ai:chat_window context{} "no-reader" line, because the member left with the def.
generated	api-surface/, export-origins/, declaration-map/, authorable-surface/ (4 lines, proof: "def no longer emitted by this build"), authorable-defaults/, json-schema.manifest/ (1 key), docs-import-surface.baseline.json, the strictness-ledger counts, and the reference docs. Each was regenerated by its generator. The manifest and authorable-surface lines were deleted as the build gate prescribes, and its proof check accepted them. authorable-surface.base.json is untouched.
docs	content/docs/ui/pages.mdx component list says the truth.
lint (comment only)	validate-page-field-bindings.ts used AIChatWindowProps.agentId as its example of a non-field string prop. The example now names a live one, ElementImagePropsSchema.alt.
changeset	@objectstack/spec minor, BREAKING banner, Clause-②: yes (narrowing), FROM → TO table, ADR-0087 marker registered ui-ai-chat-window-retired.

How this follows the precedent, and the one place the claim's wording differs

The PR #15112 shape is matched element by element: the enum value removed, a map entry, the enum error map and the node-level check sharing one string, the row kept as retiredComponentProps, pins at the three doors with a control, a vocabulary pin, regenerated artifacts, the docs list and a minor changeset. The claim and the dispatch said that "the ComponentPropsMap row ... go[es]". The precedent does not delete the row. It replaces the row's props schema with a whole-bag refusal, and its docblock says why: deleting the row would turn a loud retirement into a silent skip on every reader that dispatches on it. Two tests on this tree depend on that. component-type-vocabulary.test.ts requires every member of RETIRED_PAGE_COMPONENT_TYPES to stay isKnownComponentType. objectui's registry-inputs-spec-parity.test.ts at the pin classifies a retired type through its kept row. So the row stays and AIChatWindowProps goes. That is how "the type leaves ... its ComponentPropsMap row" lands here.

Two kit items go beyond PR #15112, because the ruling asked for them. #15112 had no D3 entry: its row had zero keys, and its disposition was not-required (no-migration-prescription). This one has a D3 entry because triage named one. It is in the element-node shape of element-filter-and-form-node-refused. It also has a RETIRED_DEFS_BY_MAJOR entry, because a published def leaves the build. No D2 conversion is registered. The precedent registered none, and the only edit is deleting an authored page node, which is a layout decision that conversions must not make.

Census (A1), at 529d9711fb

• ai:chat_window: examples/** 0, packages/apps/** 0, apps/** 0, skills/** 0. content/docs/** had 3 hits: 2 in the auto-generated references and 1 in the pages.mdx list. The other packages/** hits were the spec's own list and row, its tests, and a frozen formatter fixture in scripts/format-type.test.ts, which is left as is, as feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 left user:profile in it. Control in the same query shape: element:divider appears in 3 example files and record:details in 12.
• AIChatWindowProps: the def, the generated artifacts, and one illustrative comment in packages/lint. That comment is updated, and it is not a consumer.
• objectui at the pin 89cad75d55: there is no renderer, and the palette excludes the type. No non-test source imports AIChatWindowProps or indexes the row. The 'ai:chat_window' key sits only in Record objects keyed by string, and in comments and tests. So the Console Pin Gate's build does not reference the removed names.
• skills/** names nothing, so it is untouched (no Tier H).

Ablation (both legs committed first, mutated through scripts/ablation-replace.mjs with an anchor that must hit, restored by blob)

The subject resolves by relative import (./page.zod, ./component.zod) to src, so no dist rebuild was involved.

• Leg A, the dispatch's ablation: delete the ai:chat_window entry from RETIRED_PAGE_COMPONENT_TYPES (anchor 1 → 0, blob 7d2faf0b1f79 → 501c4251ee90). The observed direction is louder than the pin. component.test.ts fails to load, with 1 file failed and no tests run. The error reads: retiredComponentProps: ai:chat_window has no RETIRED_PAGE_COMPONENT_TYPES entry (page.zod.ts). The kept row's factory refuses a row that has lost its prescription.
• Leg B, so the by-name pins themselves judge: the same deletion, plus the row turned into a non-throwing emptyProps('ai:chat_window'). Result: 5 failed and 360 passed. The failures are the map pin, the row pin, PageComponentSchema by name, PageSchema at the element path, and the enum error map.
• Restore: both files' blobs equal HEAD (7d2faf0b1f79, 891d947a3232), git diff HEAD is empty, and git status is clean. Re-run after the restore: component.test.ts has 365 of 365 passing.

Gates, at 3f17d5229e (merged origin/main 88fb5e85a0, which touched no spec, lint or docs file this PR edits)

• @objectstack/spec build exit 0. check:generated: ✓ All 15 generated artifacts are up to date.
• @objectstack/spec test (--project local): 603 files, 17845 passed and 1 todo. typecheck exit 0, with check:test-typecheck: OK.
• @objectstack/lint build closure, then test: 119 files, 5615 passed and 5 skipped. flags ai:chat_window runs off the map. typecheck exit 0.
• check:liveness exit 0, check:migration-registry ✓ ... is current (354 semantic, 246 retired-key, 218 retired-def), check:doc-authoring exit 0, check:nul-bytes OK (... no raw ASCII control bytes).
• Changeset gates, with this body as the --event payload: check-adr-0087-registration ✓ 1 declared-breaking changeset(s) ... registered ui-ai-chat-window-retired (new here: ui-ai-chat-window-retired). check-changeset-no-major ✓ LEVEL AXIS: this PR declares clause-② yes (narrowing). check-empty-changeset exit 0.
• dispatch-gates --ran: ✓ 114 derived famil(ies) accounted for — 112 run, 2 NOT-MEASURED.
  • NOT MEASURED: check:dual-build-cjs-loads, reason: it needs a whole-repo pnpm build, and it exited 3 with PREREQUISITE NOT MET.
  • NOT MEASURED: check:type-check-debt, reason: repo-wide tsc, killed by a 420s per-gate timeout on the shared box. The two packages this PR touches typecheck green.
  • Both are declared to CI.
• NOT MEASURED: @objectstack/spec test:repo (--project repo), reason: one idle worker under 2% CPU for 10 minutes with no output, stopped. It is CI's.

Acceptance notes

• objectui will need a reconciliation when it next bumps @objectstack/spec. It is the same step objectui#7122 took for user:profile and element:form. At the pin, block-config.test.ts ("every exclusion names a real spec type") and the UNJUDGED_SPEC_BLOCKS entry in registry-inputs-spec-parity.test.ts ("NOT REGISTERED, DELIBERATELY", whose class is exclusive with "RETIRED UPSTREAM") both still treat ai:chat_window as a live enum member. They are objectui's tests against its installed spec, so they do not affect this repo's CI. Who picks it up: the next objectui spec-bump PR.
• The enumeration pin's no-reader reason kind no longer has a user. It is left as ledger vocabulary.
• The prescription names defaultAgent for what agentId was used for. objectui at the pin forwards app.defaultAgent, limited to ask / build, to the chat dock's one agent resolver (app-shell/src/layout/ChatDock.tsx:242, hooks/surfaceAgent.ts:127-128).

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 2 package(s): @objectstack/lint, @objectstack/spec, touching 10 documentable anchor(s). ⚠️ 9 changed file(s) yielded no anchor (packages/lint/src/validate-page-field-bindings.ts, packages/spec/api-surface/ui.json, packages/spec/authorable-defaults/ui.json, …), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files.

3 hand-written doc(s) NAME something this change touched and may need an implementation-accuracy re-verification:

• content/docs/kernel/cluster.mdx (via RETIRED_DEFS_BY_MAJOR (symbol, a top-level const object))
• content/docs/protocol/objectui/layout-dsl.mdx (via ComponentPropsMap (symbol, a top-level const object), PageComponentSchema (symbol, a top-level const))
• content/docs/ui/pages.mdx (via PageComponentType (symbol, a top-level const object))

⛔ 6 release-owned page(s) also name something this change touched. These are read-only:

• content/docs/releases/v15.mdx (via PageComponentSchema (symbol, a top-level const))
• content/docs/releases/v17/17-1.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-3.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-4.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-5.mdx (via ComponentPropsMap (symbol, a top-level const object))
• content/docs/releases/v17/17-6.mdx (via RETIRED_DEFS_BY_MAJOR (symbol, a top-level const object))

content/docs/releases/ is RELEASE-OWNED (AGENTS.md "Documentation Guardrails"): release
notes are written centrally at release time, and a code PR that edits them is the exact PR
that guardrail exists to stop. They are still audited — read-only. If one of them is actually
wrong, file an issue or open a dedicated docs-only PR; do not edit it here.

What this run could not see

• 9 changed file(s) yielded no anchor (packages/lint/src/validate-page-field-bindings.ts, packages/spec/api-surface/ui.json, packages/spec/authorable-defaults/ui.json, …) — pages documenting those are invisible to this run
• 8 name(s) were too generic to anchor anything (single lowercase words)
• the SDK route bridge reached 54 of 206 client-bound route-ledger rows — the other 152 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 152: 0 are remediable by widening that discovery convention (an in-repo file declares the path; the convention did not scan it); 55 are structural — on a ledger where NOT ONE row is declared in-repo, so no discovery change reaches them at any price; 97 are undecided (no in-repo declaration, on a ledger that has other in-repo registrars — absence and an unreadable spelling are not distinguishable here). The rows themselves: node scripts/docs-audit/affected-docs.mjs --bridge-coverage
• a page that states a rule by its inputs shares no identifier with the emitter that implements the rule, so an emitter-only diff cannot list it — not on this run and not on any run. Measured on fix(driver-sql): emit varchar(maxLength) for a text field a declared index keys on #11430: content/docs/protocol/objectql/types.mdx documents the text-family column mapping by the ObjectQL type names it maps FROM (text / textarea / html) while the diff changed createColumn; it went unlisted, and it was the page that diff falsified, in four places. No shared token exists to detect this on, so a rule your change carries has to be re-read by hand in the pages that restate it.
• a key NAME is not a key, so the hand re-read the line above prescribes can land on the wrong schema. The same spelling is authorable on one governed type and a [REMOVED] tombstone on another for each of active, aria, joins, objects, template, tools and version (censused on [finding] tools is a key on BOTH AgentSchema (tombstoned, dead) and SkillSchema (live, cloud-attested), so a name-based search attributes skill examples to the agent key — it produced a false stop-the-line alarm on PR #19059 #19093 over the liveness ledger's governed types, top-level keys); nothing in a search result distinguishes the two, so a grep hit on a LIVE example reads as evidence about the DEAD key. Measured on fix(spec): the agent.tools liveness row says dead — it claimed live on a key the schema tombstoned #19059: content/docs/ai/agents.mdx was reported as contradicting the agent.tools tombstone over its tools: example at :161, which is inside the defineSkill({ block opened at :155 — the page was already correct. Settle ownership by PARSING the value against both schemas, never by the name: that literal PASSES SkillSchema, and as an AgentSchema it FAILS at tools with the tombstone prescription. ⛔ These names are not the whole class — a key retired through a .strict() guidance map leaves no tombstone in the walked shape and none of them here (tool.category, live as AIToolDefinition.category).

Coarse fallback — 138 page(s) merely mention a changed package (the pre-#9192 predicate, kept for the deliberately-wide backstop): node scripts/docs-audit/affected-docs.mjs --json ad7c3518983a1bb63fd4601954ac92d055124e42 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 7846296e0a847b299ea993b44b7adb5ad58369dc — the merge of head 3f17d5229e35c592af07f771f7d1118f26e2678e into base ad7c3518983a1bb63fd4601954ac92d055124e42, which is what actions/checkout gives a pull_request run. Not the PR head.

A worktree cut from an older main holds a different content/docs, so re-deriving there can legitimately return a different list — that is a different tree, not a wrong row. To answer on the same tree:

# while this PR is open — GitHub drops the merge commit once it closes
git fetch origin 7846296e0a847b299ea993b44b7adb5ad58369dc && git checkout 7846296e0a847b299ea993b44b7adb5ad58369dc
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin ad7c3518983a1bb63fd4601954ac92d055124e42 3f17d5229e35c592af07f771f7d1118f26e2678e && git checkout -B drift-repro ad7c3518983a1bb63fd4601954ac92d055124e42 && git merge --no-ff 3f17d5229e35c592af07f771f7d1118f26e2678e

node scripts/docs-audit/affected-docs.mjs --json ad7c3518983a1bb63fd4601954ac92d055124e42

⚠️ That checkout carried uncommitted changes, so the commit above does not fully identify what was read.

Advisory only, and a precision-first one (#9192): a page is listed because it names a
symbol, wire route or SDK method this diff touched — not because it mentions a changed
package. Each row says which anchor put it there, so a wrong row is reportable rather than
merely annoying. To re-verify, run the docs-accuracy-audit workflow scoped to these files:
node scripts/docs-audit/affected-docs.mjs ad7c3518983a1bb63fd4601954ac92d055124e42 → pass the list as
args.docs, on the commit named under Which tree this was computed on.

[objectstack-fleet]

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 3f17d5229e35c592af07f771f7d1118f26e2678e
Local-runs: none

Read-only review of PR #21531 (card #21504) against the net diff on main at the head above, the card's body and every comment (ruling 5963897014, claim 5964418092, dev report 5965081737), the head's check-runs, and the precedent the ruling names: #14159 and its landing PR #15112 (97a22639b4), read at refs. Nothing was built, run or re-run.

Check-runs on the head, read at 2026-10-03T03:42Z: 32 runs — 18 success, 2 skipped, 12 in_progress, 0 failed. Success includes Check Changeset, Spec property liveness, Governed Surface Queue Guard, Build Core, Build Docs, Type Check · source gates, Type Check · debt ledger, Dogfood Verify CLI, Dogfood Regression Gate (3/3), the card/branch/single-writer claim checks and the docs checks. Skipped: Console Pin Gate (the filter job's console output did not select it — no console build input moved) and the opt-in packed-tarball smoke. Still in progress: Test Core 1–6/6, Lint & Repo Gates, Type Check · workspace, Type Check · consumer gates, Dogfood Regression Gate 1/3 and 2/3, Temporal Conformance. An in-progress gate is a reading, not a pass: those families own their own verdicts, and the landing waits on them as it did for #15112. Nothing judged below is contradicted by any completed run.

① Derived judgments

Every accept-set and public-surface change the diff implies, named right or wrong against the ruling and the precedent's actual shape.

1. PageComponentType loses 'ai:chat_window' (ui/page.zod.ts) — an accept-set narrowing on the enum parsed alone; the enum's own error map, keyed on issue.input, returns the prescription only for the ex-legal value (pinned, with ai:chat as the stranger control). Right — the precedent's element 1, the same shape.
2. PageComponentSchema.type refuses the name at the node — the existing superRefine consults RETIRED_PAGE_COMPONENT_TYPES, so the new map entry alone closes the open string arm: code: 'custom', params.retiredComponentType, the node's own path, bare and populated, and at the element path of an authored page. Right — this is the door the ruling's "⛔ de-advertising alone is not enough" demands, and the precedent's element 2; no new mechanism was needed because feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 built it.
3. ComponentPropsMap['ai:chat_window'] becomes retiredComponentProps('ai:chat_window') — z.never, refusing {} and populated bags with invalid_type and the same string. Right, and the one place the claim's wording and the landing differ. The claim and the ruling's bullet say the row "goes"; the ruling's governing sentence is "follows the user:profile precedent", and PR feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 kept the row (its component.zod.ts adds retiredComponentProps, with a docblock saying why deleting the row would demote a loud retirement to a silent skip on every reader that dispatches on it). At this head component-type-vocabulary.ts derives KNOWN from the row keys and the vocabulary test requires every map member to stay known, so the row-less reading would have gone red. The type leaves the row as a live props schema; the row stays as the refusal. Judged as the precedent's shape, not a deviation from it.
4. AIChatWindowProps leaves @objectstack/spec/ui, and ui/AIChatWindowProps leaves the published JSON Schema set — a public-surface removal the precedent never had (its row carried no def). Reflected in api-surface/, export-origins/, declaration-map/, the manifest key, four authorable-surface/ key lines, the authorable-defaults/ mode = "float" line and the docs-import baseline; registered as RETIRED_DEFS_BY_MAJOR[18] 'ui/AIChatWindowProps', beside the step-18 siblings ui/FormFieldPublicPicker and ui/Theme, which is what the manifest deletion gate prescribes. The def had zero readers (census: the only non-spec mention was an illustrative comment in packages/lint, now re-pointed; objectui at the pin imports nothing by that name). authorable-surface.base.json is correctly untouched: it is the pinned anchor for the deletion gate and still carries the earlier retired defs' lines. Right, and declared in the changeset under "What is removed from the exports".
5. Vocabulary consequences — KNOWN_COMPONENT_TYPE_CANDIDATES loses the name (derived from the map, so the suggester cannot rename an author into it), isKnownComponentType('ai:chat_window') stays true, and RESERVED_COMPONENT_TYPE_NAMESPACES is unchanged because ai:suggestion still populates ai:. Pinned with ai:suggestion as the lit control. Right — the one divergence from feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 (where user: left the reserved set) is the derivation doing its job.
6. The kit: D3 entry ui-ai-chat-window-retired under step 18, its STEP18_RATIONALE fragment, the retired-def entry, and no D2 conversion. The ruling asked for a D3 entry; feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 had none (disposition not-required), so this goes beyond the precedent on the ruling's own instruction. No conversion matches the precedent's reasoning: the only edit is deleting an authored node, which the conversion layer does not do. Step 18 is where this head's post-cut retirements already sit (translation-widget-sub-caption-retired, ui-form-layout-inline-grid-retired). The entry's replacement and the door's prescription carry one instruction, and its acceptanceCriteria names the params.retiredComponentType the door actually emits. Right. spec-changes.json and docs/protocol-upgrade-guide.md are untouched, and correctly so: at this head neither projects any step-18 entry yet (the landed siblings are absent from both), so the changeset's line saying they "follow the schema" is prose ahead of the projection, not a missing regeneration.
7. The prescription's substance — names the floating chat overlay as the entry point, gives the imperative fix, cites ADR-0049, carries no tracker number (pinned), and points agentId's intent at the app's defaultAgent. Verified against ui/app.zod.ts at the head: defaultAgent is the ADR-0063 surface-binding knob whose resolvable values are the two platform agents, ask by default and build for authoring surfaces, resolved by the assistant chat endpoint. Right.
8. ai:suggestion untouched — pinned as the control at every door. Right, per the ruling's "Not here".
9. Everything else is comment, docs or generated: the enumeration pin loses its ai:chat_window no-reader line because the row has no members; the packages/lint docblock example is re-pointed at a live prop; content/docs/ui/pages.mdx says the truth; the references and the strictness counts are regenerated. scripts/format-type.test.ts's frozen fixture still lists the name exactly as feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 left user:profile in it, and the dated 2026-06 liveness audit still lists it as aspirational as it does element:filter / element:form — records, not drift. Right.
10. One residual the review found (not a dev flag): the KNOWN_COMPONENT_TYPES docblock in component-type-vocabulary.ts enumerates the kept retired rows as user:profile plus element:filter / element:form; ai:chat_window now belongs to that list, so the sentence is one member short. Comment-only, no publish change — the same class feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112's review carried as a report line (its deviation 2). Carried as a note for the next PR that edits that file; not a fail item.

② Semver level

Changeset .changeset/ai-chat-window-retired.md: @objectstack/spec minor, the BREAKING banner, Clause-②: yes (narrowing), a FROM → TO table, and the adr-0087 marker comment reading registered ui-ai-chat-window-retired. What the diff publishes: an accept-set narrowing at three doors (①1–3) and a public export removal (①4). Under strict semver both grade major; under the launch-window convention check-changeset-no-major.mjs enforces until GA, a breaking change ships as minor with the banner as the carrier — the level #15112 shipped at, and the level the ui/Theme and ui/FormFieldPublicPicker def retirements shipped at. The Check Changeset run on this head is success (the no-major guard, and the level axis: clause ② declared, @objectstack/spec graded minor). Level: right. The Clause-②: line reads yes (narrowing) in the PR body, the changeset and the claim, and the content is a narrowing (an authored type refused where it parsed clean before). Right. The ADR-0087 disposition is registered, not the precedent's not-required, because a D3 entry exists here. Right.

③ Boundary flags

Dev flags (eight deviations, two out-of-scope findings, no open questions), each answered:

• Row kept, not deleted — answered at ①3: the precedent's shape; the ruling's "follows the precedent" governs over the claim's paraphrase. Accepted.
• Kit beyond feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 (D3 + retired-def) — ①6: ruling-mandated and gate-mandated. Accepted.
• Files outside the claim surface (packages/lint comment, pages.mdx, strictness counts) — a comment, a hand-written doc line and a generated count; no publish change. Accepted.
• Prescription names defaultAgent — ①7, verified on the spec side at this head. Accepted.
• origin/main merged before the PR opened — the review is of the net diff against main, which the merge does not change. Accepted.
• test:repo not measured locally — CI's Test Core shards own it on this head and are in progress; not a contract item. Read as it stands.
• Trailer and footer forms — repo convention, not a contract item.
• Out of scope 1: objectui will need reconciling at its next spec bump (block-config.test.ts and registry-inputs-spec-parity.test.ts at pin 89cad75d55 still treat the type as a live member). Accepted as a carried note with its carrier named (the next objectui spec-bump PR, as objectui#7122 did). A recommendation to the seat, not a ruling: post the consumer-side facts to objectui at landing, as the Ruling needed: does user:profile get a renderer, or become explicitly not author-placeable? — the one #12183 sibling that ruling never covered (gates objectui#7135) #14159 landing did for its lane.
• Out of scope 2: the no-reader reason kind is now unused vocabulary — carrier spec(ui): the ComponentPropsMap rows still type renderer-read members as z.unknown() — navigation on object-map / object-gantt / object-tree and conditionalFormatting on object-kanban accept 42 — the family close-out after #21445 #21464's next stage. Accepted.
• Open questions: none declared, none found. The one reviewer-found residual is ①10, a report line.

Implemented-by: claude/issue-21504-retire-ai-chat-window
Reviewed-by: session_01YDt3PzwfrkuFzUBF89WPmM

VERDICT: PASS

---

Generated by Claude Code

[objectstack-fleet]

ACCEPT — PR #21531 @ 3f17d522 (#21504)

domain:spec seat 2 (session_01YDt3PzwfrkuFzUBF89WPmM), holder of claim 5964418092 · 2026-10-03T03:55Z

• Shape (read on GitHub): a draft against main. The first line is Fixes #21504, and Clause-②: yes (narrowing) sits at a line start. 22 files, +393 / −74, most of them generated.
• Review: the contract review at CONTRACT_REVIEW_TIER, 5965171663, reads PASS on this head.
  • Retired at three doors: 'ai:chat_window' leaves PageComponentType, and the page node refuses it by name through RETIRED_PAGE_COMPONENT_TYPES. The refusal message names the floating chat overlay, points agentId's intent at the app's defaultAgent, and carries no tracker number.
  • The props bag is refused too: the ComponentPropsMap row becomes retiredComponentProps, which refuses every bag.
  • AIChatWindowProps leaves the public surface. It is registered in RETIRED_DEFS_BY_MAJOR[18], and the D3 entry ui-ai-chat-window-retired sits under step 18. There is no D2 conversion, as in the precedent.
  • ai:suggestion is untouched and is the pinned control.
  • Release: @objectstack/spec minor with the BREAKING banner and a FROM → TO table. The ADR-0087 marker reads registered.
• The row is kept, and this seat accepts it. The claim said the ComponentPropsMap row "goes". The ruling's governing sentence is "follows the user:profile precedent", and PR feat(spec): user:profile is explicitly not author-placeable — refused by name at the schema door (#14159) #15112 kept the row as the refusal. The vocabulary test at this base also requires every retired type to stay known. The claim's wording was this seat's paraphrase, and the precedent's shape governs. The review reads the row the same way (①3).
• CI on 3f17d522: 33 success and 2 skipped by design (Console Pin Gate, packed-tarball smoke).
• Governed surface: check-governed-merges --pr 21531 reads NOT governed: 0 of 22 paths, 467 changed lines.
• Serial: at this read, no other open PR touches ui/page.zod.ts, ui/component.zod.ts, the enumeration pin or migrations/registry.ts. [Decision] page requires on the page kinds whose source is never compiled at save (react, full, slotted): refuse it at parse, or keep admitting a key nothing derives? #21459 (ruled A, PageSchema.requires) also writes ui/page.zod.ts and the step-18 rationale. Its claim names this PR, and its dev merges main after this lands.

Carried, decided here:

• objectui, at its next @objectstack/spec bump (the one that resolves a release carrying this changeset, as objectui#7122 did for user:profile): two objectui tests still treat ai:chat_window as a live spec member, at the .objectui-sha pin 89cad75d55. They are packages/app-shell/src/views/metadata-admin/previews/__tests__/block-config.test.ts (PALETTE_EXCLUSIONS, "every exclusion names a real spec type") and apps/console/src/__tests__/registry-inputs-spec-parity.test.ts (UNJUDGED_SPEC_BLOCKS, to move from "not registered, deliberately" to "retired upstream"). Both fail loudly in that bump's own CI. → noted on the card at landing, not filed.
• The KNOWN_COMPONENT_TYPES docblock in component-type-vocabulary.ts lists the kept retired rows and is now one member short (review ①10, comment only). → It rides spec(ui): the ComponentPropsMap rows still type renderer-read members as z.unknown() — navigation on object-map / object-gantt / object-tree and conditionalFormatting on object-kanban accept 42 — the family close-out after #21445 #21464's next stage, which edits that region's neighbour, the enumeration pin.
• The pin's no-reader reason kind has no user now. → Also spec(ui): the ComponentPropsMap rows still type renderer-read members as z.unknown() — navigation on object-map / object-gantt / object-tree and conditionalFormatting on object-kanban accept 42 — the family close-out after #21445 #21464's next stage.

Next: ready, auto-merge, the queue.

---

Generated by Claude Code