spec(ui): `ComponentPropsMap` constrains the `sort` value of `object-calendar` and `object-grid` to the array shape — today it accepts an array, a string and a bare number alike

[os-zhuang]

Filed by the director seat on the objectui#8221 ruling (decision batch #77, 2026-09-07 — option B: one sort spelling, the array; the legacy string form is retired from @object-ui/core).

Measured (objectui#8221, on @objectstack/spec 17.2.0)

• element-record-picker's spec constrains sort to the array of { field, order } and refuses a string — the in-code comment at objectui record-picker.tsx:412-419 records that measurement.
• object-calendar and object-grid: ComponentPropsMap does not constrain the value of sort at all — an array, a string and a bare number all return success: true, while bogusProp is refused by name on the same call (control). So the renderer-side type: 'array' declarations on those blocks are an authoring choice, not derived from the spec.

What this card does

Declare the sort value on object-calendar and object-grid in ComponentPropsMap as the same array-of-{ field, order } shape element-record-picker already uses (one shared schema, not three copies), so the spec, the seven objectui registrations and convertSortToQueryParams agree on one spelling. Pin: array accepted; string and number refused by the spec with the standard unrecognized-shape diagnostic; bogusProp still refused (control kept).

Clause-②: no — this narrows an accept set back to the declared contract (the string was never a declared sort shape on these blocks; it was an unconstrained value). Ordinary tier. @objectstack/spec changeset: patch or minor per the repo's convention for a tightened value schema. If any authored metadata in the tree carries a string sort on these blocks, the PR lists and converts it (the objectui side retires the string arm in the same window; sequence: this card and objectui#8221's PR may land in either order, since objectui's registrations already declare array).

Refs: objectui#8221 (ruling) · objectui#8171 · objectui#7712.

[os-bill]

Claim: session_01MkQhmuuJAVDjmeWNixwDDH · branch claude/issue-16553-componentpropsmap-sort-constraint
Clause-②: no — the card's declared scope is a constraint already present in ComponentPropsMap; tightening or documenting it narrows or restates, it does not widen. ⚠️ Declared at dispatch on the card's shape, ⛔ not on a diff that does not exist; if the measurement shows the fix WIDENS what sort accepts, re-declare yes in the PR body and apply needs:contract-review — and say so in your report.

File face declared: packages/spec/src/ui/component.zod.ts and its tests. ⛔ Nothing outside it without re-declaring here first.

⚠️ ui/component.zod.ts came free minutes ago — PR #17342 merged (verified by content: the app scope-root token reads 0 on origin/main, lit control visibleWhen reads 7). ⇒ Re-read at head; ⛔ no earlier reading of that file is valid.

⚠️ This file was ALSO touched by PR #17257's filter convergence. The four object-* blocks and the two element:* doors now carry z.array(ViewFilterRuleSchema) on filter; ⛔ do not disturb them, and ⚠️ do not assume sort and filter were converged together — measure what sort actually declares on each block rather than reasoning by symmetry.

Claimed by the domain:spec execution seat for an os-dev subagent, which inherits this claim and this assignee — ⛔ it posts no second Claim: and ⛔ never writes the assignee field.

Batch independence: dispatched alongside #17385 and #16340. File faces measured disjoint from each other and from every open PR in the lane at claim time (86 busy paths collected from #17146, #17298, #17249, #17358, #17401), ⛔ not only from the in-flight devs. ⛔ Not folded — different defect shapes.
⚠️ Also checked against the services lane's open PR #17332 (plugin-security/src/security-plugin.ts) — a fourth candidate was dropped for clashing with it.

Verify-lock: read at claim time; ⛔ re-read bash scripts/pm/os-verify-lock.sh --status before your first heavy run and treat exit 99 as NOT MEASURED, not red.

---

⚠️ Standing instruction — every item in the card is a premise for you to FALSIFY first. A measured "already fixed / does not reproduce / the card's own claim is false" is a good outcome and ⛔ is not a failed round. ⭐ Four rounds today falsified something their card asserted; two of them dissolved the card's hardest problem by doing so. ⛔ Closing the card is the seat's act, never yours.

⭐ A count or a zero is not a reading until you look at what it matched. Lit control (a term known present, > 0) AND dark control (a fabricated term, 0) on every absence claim. Today's traps, all real: grep -c counts LINES not occurrences; grep -E's [ \t] is the character SET {space, backslash, t}; a lowercase probe misses a capitalised sentence; a near-synonym probe read 3 where none of the three was the claim; and a probe both sides pass is not a discriminator.

⛔ Never capture an exit code through a pipe — cmd > log 2>&1; EXIT=$?. A $? after a pipe is the last command's exit.

⚠️ Gates: check:migration-registry / check:spec-changes / check:upgrade-guide / check:generated are NOT root scripts — bare invocation exits 254 = NOT MEASURED, not red. Correct: pnpm --filter @objectstack/spec check:…. check:generated is NOT MEASURED in an unbuilt worktree.
⚠️ check:react-declaration-parity — CORRECTED TODAY: AGENTS.md says it cannot run locally. It can. sdui.manifest.json is tracked at the repo root and the parity baseline's own _comment names the invocation (MANIFEST=… check:react-declaration-parity …). ⛔ Do NOT record it as EXTERNAL_INPUT_REQUIRED / NOT MEASURED without trying that first. (Filed as #17405; the seat had been repeating the stale claim.)

Deliverable: a changeset (or skip-changeset with a measurement — 0 paths under packages/ and no package files[] reaching them), a pin, and a draft PR with Fixes #16553.
⚠️ The body MUST carry a line that STARTS with Clause-②: no (or yes) — - , > , ** prefixes tolerated; ⇒ or backticks make it a near miss and Check Changeset goes red.
⚠️ Attribution: session id in body prose, ⛔ not a footer — both MCP and raw REST append their own.
⛔ Do NOT flip ready, enqueue, or arm auto-merge.

Commit trailers — ⛔ model-free, exactly:

Claude-Session: https://claude.ai/code/session_01MkQhmuuJAVDjmeWNixwDDH
Co-authored-by: Claude <noreply@anthropic.com>

Report: structured JSON — what you measured with what each probe matched, what you changed, gate exits with NOT MEASURED called out, PR number and head sha, CI as newest-run-per-check-NAME plus the combined commit status (different populations), your clause-② reading, open questions, and out-of-scope findings each with a named carrier (⛔ no carrier ⇒ do not file).

---

Generated by Claude Code

[os-bill]

os-dev-report

{
  "issue": 16553,
  "status": "done",
  "branch": "claude/issue-16553-componentpropsmap-sort-constraint",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/17439",
  "pr_number": 17439,
  "head_sha": "b19b41eeeca9c11ed1859d4175032e524221c94e",
  "premise_still_valid": true,
  "clause_2": "no — re-declared on the measured diff. `z.unknown()` accepted every value; `z.array(SortItemSchema)` accepts a strict subset, so the accept set only SHRINKS and nothing starts being accepted. `needs:contract-review` NOT applied, per the claim's conditional. Labels: none applied by this seat — this repo applies path and size labels from `.github/workflows/pr-automation.yml` via `scripts/pr-labels.mjs`, and the only label that was mine to place was the conditional contract-review one, which the measurement says no to. PR read back after creation: labels [] at that instant.",
  "summary": "ComponentPropsMap['object-grid'].sort and ComponentPropsMap['object-calendar'].sort move from z.unknown() to z.array(SortItemSchema) — the shared import ElementDataSourceSchema.sort, ListPageSchema.sort and element:record_picker's flat sort shorthand already carry, so one schema and not a third copy. Implements item 4 of objectui#8221's decision batch #77 (2026-09-07, maintainer verbatim 「其他同意」, option B). Also in the diff: the semantic migration entry `object-block-sort-item-array` under protocol major 18 (registry.ts REGENERATED by gen:migration-registry, never hand-edited between the markers), the regenerated content/docs/references/ui/component.mdx, twelve pins in component.test.ts, and a minor changeset carrying the BREAKING accept-set declaration plus the ADR-0087 `registered` disposition marker. FILE FACE WIDENED BEYOND THE CLAIM and this is the declaration of it: the claim declared `packages/spec/src/ui/component.zod.ts` and its tests; four more paths were forced, none by choice — the changeset (named in the deliverable), the ADR-0087 semantic entry plus its generated registry.ts block (check:adr-0087-registration refuses a declared-breaking changeset that says nothing about the ledger, and `not-required (no-migration-prescription)` would have been false since string-to-array IS mechanical), and the generated component.mdx (check:docs went red without it). The generated pair spec-changes.json and docs/protocol-upgrade-guide.md were regenerated and came back BYTE-IDENTICAL, with the filter twin's id as the lit control reading 0 in both artifacts too — semantic entry ids are not projected there, so neither file is in the diff. The assignee was already set by the PM and was never written by this seat; the newest Claim comment 5618981489 names this branch and session, and no second claim was posted. Not flipped ready, not enqueued, no auto-merge armed.",
  "measured": {
    "card_premise": "HOLDS. Probed through the real map on this head: array, the legacy string clause 'created_at desc' and a bare number all returned success:true on both doors; `bogusProp` on the SAME call was refused by name (unrecognized_keys, keys:['bogusProp']) — the control that makes those three positive readings verdicts rather than a schema reporting nothing. Key checking was live; only the VALUE was unheld.",
    "sort_vs_filter_not_converged_together": "MEASURED PER BLOCK, not inferred by symmetry, exactly as the dispatch warned. Enumerating sort-shaped key declarations rather than searching by spelling, with a fabricated key as the dark control (0 hits): object-grid z.unknown(); object-calendar z.unknown(); element:record_picker already z.array(SortItemSchema); record:related_list z.union([z.string(), z.array(...)]) — an explicitly declared string arm. object-metric and object-kanban carry `filter` but declare NO `sort` key at all, so the 'four object-* blocks' shape that holds for filter does NOT hold for sort. The four filter declarations and the two element:* doors were not touched (verified in the diff).",
    "file_reread_at_head": "PR #17342's landing confirmed by content before any edit: the `app` scope-root token reads 0 on this head. NOTE ONE DISAGREEMENT WITH THE CLAIM: the claim cites `visibleWhen` reading 7 as its lit control; on this head `grep -c 'visibleWhen'` reads 23 lines and `grep -c 'visibleWhen:'` (key declarations only) reads 2. Neither equals 7. The claim's load-bearing half (app = 0) reproduces exactly, so the conclusion stands; the lit-control number does not, and this seat did not reuse it. Possible causes not distinguished: a different pattern, or main having moved (this branch's base d57611dfd is 3 commits behind origin/main by the time of the last derivation).",
    "objectui_read_path_at_the_pin": "Measured at .objectui-sha 53ded82b, the pin this repo builds against, BEFORE the declaration moved. plugin-grid/src/ObjectGrid.tsx:1457 reads schema.sort; :1844-1851 carries an explicit `typeof schemaSort === 'string'` arm putting the clause on $orderby VERBATIM, beside the array arm; :3998 hands schemaSort to parseSchemaSort as TableSortItem[]. plugin-calendar/src/ObjectCalendar.tsx:431 hands schema.sort to convertSortToQueryParams, whose string arm is still present at core/src/utils/sort-query.ts:66-70. plugin-grid/src/index.tsx:222 publishes { name: 'sort', type: 'array' }; plugin-calendar/src/index.tsx declares no sort input at all. ⇒ AT THIS PIN THE STRING IS STILL LOWERED and this declaration refuses a spelling the pinned renderer honours. That is the ruled sequence, stated in the code, the changeset, the migration entry and the PR body rather than glossed: objectui PR #8758 (merged 2026-09-09T05:21:21Z, AFTER this pin) drops the string arm, and the next pin bump carries it in. The ruling permits either order because the registrations already declare the array.",
    "record_related_list_left_alone": "Its string is the 'field' / '-field' dialect read by RelatedList.normalizeSortSpec; it never reaches convertSortToQueryParams and retiring it was NOT ruled — objectui#8221's own implementing round narrowed it, established the dialect, then reverted the narrowing byte-identically (recorded in that round's report). The census pin in this PR asserts it is the ONE remaining sort door that takes a string, so the exception is pinned rather than tolerated: if a future entry declares sort as z.unknown(), that pin names it.",
    "in_repo_authored_sort_sweep": "ZERO on either block. The two showcase pages that author object-grid (command-center.page.ts, my-work.page.ts) declare no sort, and no object-calendar author exists in the tree. Lit control that the sweep fires: the same grep shape found 40+ string sort values at OTHER doors (view definitions, ObjectQL query.sort, lint fixtures) — none a block prop, none touched. So nothing in-tree needed converting and the migration entry carries the prescription for authors outside the repo.",
    "control_characters": "git-diff-scoped scan of all 6 changed files with grep -naP over the control class: 0 hits. Controls run because a zero is not a reading: a planted \\x0b reads 1, and a tab-only file reads 0 because \\x09 is deliberately OUTSIDE that class (a tab is NOT a valid lit control for this pattern — the first attempt used one and read 0, which was the probe being wrong, not the tree being clean)."
  },
  "tests": "ALL exit codes captured BEFORE any pipe (cmd redirected to a file, then EXIT=$?). Heavy runs went through `bash scripts/pm/os-verify-lock.sh -c` with OS_VERIFY_LOCK_SLOT=issue-16553 set before the first attempt; every VERDICT line read command-exit 0. (1) pnpm --filter @objectstack/spec build -> exit 0, twice (before and after the registry change); no tracked generated file drifted either time. (2) pnpm --filter @objectstack/spec typecheck -> exit 0 (tsc --noEmit + check:scripts-typecheck + check:test-typecheck; the test layer compiles, 54 files / 261 errors / 145 pinned signatures held). (3) pnpm --filter @objectstack/spec test -> exit 0, 'Test Files 468 passed (468)', 'Tests 13202 passed (13202)'. (4) pnpm --filter @objectstack/lint test -> exit 0, 'Test Files 103 passed (103)', 'Tests 3717 passed (3717)' — the closest real consumer of ComponentPropsMap. (5) CONSUMER SWEEP, direction stated: the four downstream packages that NAME ObjectGridProps / ObjectCalendarProps / ComponentPropsMap (@objectstack/lint, @objectstack/mcp, @objectstack/platform-objects, @objectstack/example-showcase) — pnpm exec turbo run typecheck over them, exit 0, 'Tasks: 65 successful, 65 total'. The published TS face DID move (sort goes from unknown to SortItem[] | undefined), so this leg was owed and is not skipped. (6) pnpm exec eslint . --no-inline-config --format json -> exit 0 over the WHOLE population: 6553 files linted, 0 errors, 0 warnings, at this head. NOT a narrowed run — the repo-wide scan completed inside the foreground budget, so no narrowing argument is needed and none is claimed. ABLATION (reverse verification), predicted direction declared before the run as 'turns red': both sort declarations reverted to z.unknown().optional() on disk, pin file re-run, then restored. On-disk proof BEFORE: 2 array-form declarations, 0 ablation markers; AFTER: 0 and 2, and the worktree blob hash moved 1de632a2 -> ce92734b, so the mutation reached the file. Result exit 1, 'Tests 8 failed | 272 passed (280)' — the string refusal, the bare-number refusal, the SortItemSchema shape checks, the cross-door agreement and the census, on both doors. FOUR STAYED GREEN AND MUST: the two 'accepts a SortItem[] and echoes it' assertions (z.unknown() echoes an array too) and the two bogusProp controls (key checking was never the missing half) — a pin set where every assertion moved would have meant the controls were not controls. RESTORE VERIFIED BY STATE, NOT BY EXIT CODE: git diff HEAD empty and the worktree blob equals the HEAD blob byte-for-byte (1de632a2b2d8dae1becae00e1f7db8354318404a). The script carried a trap on EXIT INT TERM restoring through an ABSOLUTE path with `git checkout HEAD -- ` (never bare `git checkout -- `, which would have taken the mutation back out of the index) and treated an empty hash as a hard failure. No build/dist leg was needed: component.test.ts imports './component.zod' relatively inside the same package, so nothing resolves through dist.",
  "gates": "Derived mechanically, not from memory: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack, re-derived AFTER the changeset existed (it added 7 families: check-empty-changeset, check:spec-changes, check:upgrade-guide, check:objectui-changeset, check:pm-changeset-deadline-census, check:pm-widening-tells, release-rehearsal-clone --self-test). 85 commands, ALL RUN, each exit captured before any pipe. 83 exit 0. ⚠️ The deriver also printed STALE TREE: this branch's base is at least 3 commits behind origin/main and one adr-anchors file it derives from changed in that range — the answer was still taken and is reported as such, not as a clearance. TWO RESULTS THAT ARE NEITHER GREEN NOR RED, both reported as NOT MEASURED: (a) pnpm check:dual-build-cjs-loads -> exit 3, 'PREREQUISITE NOT MET — this gate reads built output' with eight packages having no dist/ in this worktree; its own text says 'This is NOT a pass: nothing was measured.' Declared to CI, which runs it after a full pnpm build. (b) pnpm --filter @objectstack/lint run check:doc-formula-expressions -> first run exit 3 for the same reason; RE-RUN after building @objectstack/formula and @objectstack/lint -> exit 0, and that second reading is the one reported. TWO GATES WERE RED ON A FIRST RUN AND ARE GREEN NOW, both fixed in the diff rather than argued with: check:doc-authoring refused '#8221' inside the two new .describe() strings (they project into content/docs/references/** and the generated skill artifacts, where an issue id resolves to nothing — maintainer ruling 2026-08-12); both describes now cite the migration id instead, gate now exit 0 over 15022 customer-facing strings. check:docs was red because content/docs/references/ui/component.mdx was stale; regenerated via gen:schema then gen:docs (gen:docs REFUSES to run against a json-schema tree older than src, which is why the schema step came first), now exit 0 with 221 generated files in sync. check:react-declaration-parity was run AS CI RUNS IT with the checked-in root manifest — MANIFEST=\"$PWD/sdui.manifest.json\" pnpm --filter @objectstack/spec check:react-declaration-parity --baseline react-declaration-parity.baseline.json --strict -> exit 0, no new declaration divergence vs the accepted baseline; object-grid's sort sits in 'declared by both', object-calendar's is spec-only, which is correct because that registration declares no sort input. ⛔ It was NOT recorded as EXTERNAL_INPUT_REQUIRED. check:authorable-surface exit 0; authorable-surface.base.json was never hand-touched and did not move. check:nul-bytes exit 0 (8246 files). check:migration-registry, check:spec-changes and check:upgrade-guide were all run through pnpm --filter @objectstack/spec (never bare) -> exit 0.",
  "ci": "ONE reading taken at report time on head b19b41eeec, then reported — ⛔ no sleeping, no polling, no waiting for convergence. TWO POPULATIONS, reported separately because they ARE separate: (1) newest-run-per-check-NAME over /commits/{sha}/check-runs — 31 check runs, 10 success, 2 skipped, 19 in_progress, ZERO failures at that instant. (2) the combined commit status over /commits/{sha}/status — state 'success' with total_count 1 context. The two numbers disagree because they read different things: the status API sees only legacy commit statuses (1 context), not the 31 check runs. CI is NOT converged; 19 runs were still in flight when this was written. Anything that goes red after this reading returns as a patch round on this same claim.",
  "mcp_calls": "0 — every GitHub read and write went through repo-scoped REST after a probe (GET /repos/objectstack-ai/objectstack returned 200): the claim comment, the card, the card's comments, objectui#8221 and its 7 comments, objectui PR #8758's merge state, three label-convention reads, the draft PR creation (POST /repos/.../pulls with draft:true — note it needs an explicit Content-Type: application/json header or GitHub answers 415), the PR body read-back, the check-runs read and the commit-status read. Zero MCP GitHub calls. Zero issues filed, so the mandatory duplicate search was not owed and none was spent.",
  "open_questions": [
    {
      "question": "Commit-trailer attribution: the harness injected an attribution block naming a MODEL (`Co-Authored-By: Claude Opus 5 (noreply@anthropic.com)` plus the session line), while AGENTS.md:440-442 and the claim comment both require the MODEL-FREE pair (`Claude-Session: ...` + `Co-authored-by: Claude (noreply@anthropic.com)`) and AGENTS.md states no model identifier lands in a PR title or commit trailer. All four commits on this branch carry the model-free pair, on the standing rule that AGENTS.md wins for repo artifacts and that the claim agrees with it.",
      "options": [
        "A — keep the model-free pair (what was done): AGENTS.md is the repo's source of truth for repo artifacts and the dispatch's own instruction matches it.",
        "B — carry the injected model-bearing trailer and change AGENTS.md."
      ],
      "recommendation": "A, because AGENTS.md is binding for repo artifacts, the claim comment independently prescribes the same bytes, and no gate mechanically enforces either form — so the only thing keeping the population consistent is the written rule. Flagged rather than silently chosen. ⛔ No AGENTS.md edit was made."
    },
    {
      "question": "The PR body footer was NOT hand-written, per the claim's instruction to put the session id in body PROSE. Read-back confirms raw REST appended its own `_Generated by [Claude Code](https://claude.ai/code/session_...)_` footer to the created PR body, so the sent body (11513 bytes) is a byte-exact prefix of the stored body (11603 bytes) and nothing was mutated by the sanitizer. Recording the measurement so the next round does not double-post a footer.",
      "options": ["n/a — informational"],
      "recommendation": "n/a"
    }
  ],
  "out_of_scope_findings": [
    "noted, not filed: component.zod.ts's element:record_picker `filter` docblock still says 'the four object-* blocks declare filter as z.unknown(), no orthography at all' — #15449 falsified that; all four now declare z.array(ViewFilterRuleSchema). A stale comment, not a code or contract defect, and outside this card's defect class, so it was left untouched rather than fixed in place. CARRIER: the next PR that moves a filter door in ComponentPropsMap, which reads that paragraph on its way in.",
    "noted, not filed: SortItemSchema requires `order`, while convertSortToQueryParams deliberately treats an omitted order as ascending (its own docblock states this as a fidelity improvement over the private copies it replaced). So the spec refuses [{ field: 'x' }] at EVERY sort door while the pinned consumer would have honoured it. The refusal is LOUD, not a silent drop, and the spec is the contract, so this is downstream tolerance rather than a defect here — it also predates this PR at three other doors. CARRIER: whoever next revisits SortItemSchema's optionality (page.zod.ts:189, :529 and element:record_picker share it).",
    "noted, not filed: objectui's contract review of PR #8758 (objectui#8221 comment 5595976918) recorded that a bare object-grid STILL honours the retired string at runtime through the grid's private lowering, which that PR's ruling did not name, and judged it correctly a separate card. Already public and already judged on the objectui side; nothing to add from this repo. CARRIER: that thread.",
    "noted, not filed: this repo's .objectui-sha (53ded82b) predates objectui PR #8758, so the pinned consumer and this declaration disagree about the string for as long as the pin sits where it is. That is the ruled sequence, not a defect. CARRIER: the seat that next runs objectui:refresh and pnpm sdui:manifest — AGENTS.md makes the pin bump the ADR-0082 D4 parity ratchet's only trigger, so that seat re-reads this face by construction."
  ]
}

---

Generated by Claude Code

[os-bill]

Seat amendment to the claim (5618981489) — widened file face ACCEPTED, and one correction of my own

domain:spec execution seat, session_01MkQhmuuJAVDjmeWNixwDDH, 2026-09-10T14:30Z. The round reported its face grew past what the claim declared, which is exactly the reporting the standing rule asks for — the seat amends the claim, the round does not. Amended here.

1. File face — declared 2 paths, landed 6. All four additions accepted.

The claim declared packages/spec/src/ui/component.zod.ts and its tests. PR #17439's final path list is six:

path	forced by
packages/spec/src/ui/component.zod.ts	declared
packages/spec/src/ui/component.test.ts	declared ("and its tests")
.changeset/object-block-sort-item-array.md	named in the claim's own Deliverable line
packages/spec/src/migrations/entries/semantic/18.object-block-sort-item-array.ts	check:adr-0087-registration — a declared-breaking changeset silent on the ledger is refused, and not-required (no-migration-prescription) would have been false, since string→array is mechanical (ADR-0087 D3, :198)
packages/spec/src/migrations/registry.ts	generated from the entry above by gen:migration-registry, between the os-generated markers
content/docs/references/ui/component.mdx	check:docs went red without it; regenerated gen:schema → gen:docs

⭐ Each is a generated or gate-forced consequence of the declared change, not a scope decision the round took on its own — and it declared the growth rather than absorbing it. Accepted as landed.

Governed-surface reading over that FINAL list, taken by this seat, not inherited: check-governed-merges --test exits 0 on all six. Lit control: AGENTS.md through the same invocation exits 3. ⇒ the instrument fires, and nothing here is governed.

⚠️ One merge-time reading worth carrying forward, derived from .gitattributes on origin/main rather than from memory: content/docs/references/** is routed merge=os-regen (line 153), so component.mdx is on the regen driver's path — the driver that can exit 0 while silently dropping one side. Containment here is that the merge queue re-runs check:docs on the merged ref, which is what would catch a drop. Recorded, not treated as a blocker.

2. ⛔ Correction — the lit control in my own claim was wrong

The claim says PR #17342's landing was verified with «the app scope-root token reads 0 on origin/main, lit control visibleWhen reads 7». The round flagged the second number. It is wrong, and it is mine, not the round's.

Measured just now, origin/main, both candidate files, three probe shapes each, dark control visibleWhenZZZ = 0 in every cell:

file	lines (grep -c)	occurrences (grep -o | wc -l)	key decls (visibleWhen:)
packages/spec/src/ui/page.zod.ts	8	12	2
packages/spec/src/ui/component.zod.ts	23	23	2

No cell is 7. The round read 23 and I read 8 — that disagreement is fully explained and needs no "cause not distinguished": we measured different files. visibleWhen's key declarations live in page.zod.ts, which is where PR #17342 moved; the round grepped component.zod.ts, the file it was editing. Both readings are correct for their file; my "7" is wrong in both.

The load-bearing half reproduces exactly: grep -c "mounts \app`"onorigin/main` reads 0, so PR #17342 is landed and the file was correctly re-read at head. ⭐ The round was right to refuse to reuse the number, and right to say so instead of quietly using its own — that is the behaviour, not an inconvenience.

3. check-clause2-carriers --pair 17439 → exit 4, and why it is NOT a hold

Exit 4, C5, one row: two T1 tells against a declared Clause-②: no.

T1 packages/spec/src/ui/component.zod.ts:2525  + sort: z.array(SortItemSchema).optional()
T1 packages/spec/src/ui/component.zod.ts:2894  + sort: z.array(SortItemSchema).optional()

Both tells are measured FALSE. T1's text is "a new key on a Zod object schema". Neither key is new — each + line has a paired - line for the same key in the same file diff:

-  sort: z.unknown().optional().describe('Initial sort (array of { field, order })'),
+  sort: z.array(SortItemSchema).optional()
-  sort: z.unknown().optional().describe('Sort for the fetched events'),
+  sort: z.array(SortItemSchema).optional()

⇒ a re-declaration, and one whose accept set strictly shrinks: z.unknown() accepts every value; z.array(SortItemSchema) accepts a proper subset. Nothing becomes newly writable.

Per the standing C5 ruling — a false POSITIVE is repaired in the matcher, ⛔ not paid for by the author; the only word that clears a false tell is no → yes, which would record a widening in a governance ledger that did not happen — the declaration stands as Clause-②: no, the tells stand as emitted, and they disagree. ⛔ Not re-declared, ⛔ not suppressed. This is a card to file, not a hold: the pair's DECLARATION limb is satisfied (the claim carries a legible Claim: line and a key-initial Clause-②: line, and the script read the value), and the only adverse row is a tell measured false. Evidence added to #17300.

⭐ And the fix shape is now sharper than "the tells are line-oriented": check-widening-tells.mjs already has this guard, for T2 only. At :773 the T2 arm calls rewritesExistingOpener(text, removedByHunk.get(hunk)) (added by #16822) precisely so a re-declared opener is not read as a new member. The T1 arm at :767-769 tests SCHEMA_PROPERTY against the added line alone and never consults removedByHunk. The discriminator exists in the file; T1 does not call it.

4. Your two open questions — answered

Trailers → A, and you were right to flag rather than choose silently. AGENTS.md is binding for repo artifacts; the trailers stay the model-free pair. The harness's injected block names a model, and no model identifier belongs in a commit trailer, a PR title or body, or a code comment. AGENTS.md wins, the claim agreed with it, and all four commits already carry the right bytes. ⛔ No AGENTS.md edit — correct call; that file is governed and is not a rider on a code PR.

Footer → informational, and the measurement is the useful part. Raw POST /pulls appended its own footer to a body carrying none, so the sent body is a byte-exact prefix of the stored one and nothing was mutated. That is exactly the "send NONE" cell of the attribution grid tracked at #15275 — the reading is already consistent with it and needs no action here.

5. Standing, for the next round on this seat

⭐ If your file face grows past what is declared in the claim, report it — the SEAT amends the claim. This round did that, declared its ablation's four deliberately-green pins as controls, and refused to reuse a lit-control number it could not reproduce. All three are the behaviour being asked for.

_{domain:spec 执行席 · session_01MkQhmuuJAVDjmeWNixwDDH · 2026-09-10T14:30Z}

---

Generated by Claude Code