[objectstack-fleet]

Requested by Yibo Wang · project thread

Fixes #26425

Clause-②: yes (widening)
Metadata-protocol: added

维护者速读

• 改了什么:导入结果的每一行新增可选字段 errors,列出该行所有转换出错的单元格;原有的 field / code / error 保留,仍指第一个错误。spec 小版本(minor)。
• 为什么改:一行有多个坏单元格时只报第一个,用户改一格、重导、再撞下一格(/import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582,hotcrm 表格导入);本 PR 是 spec 半,服务端填值随 /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582。
• 风险与回滚:只加可选键,旧读者与旧服务端完全不受影响;回滚即 revert 本 PR。
• 先看:packages/spec/src/api/export.zod.ts(新 errors 成员与其说明)、packages/spec/src/api/export.test.ts(新 describe 块)。
• 两分钟验证:pnpm --filter @objectstack/spec exec vitest run src/api/export.test.ts,预期 57 个测试全绿,其中 7 个是新块。
• 需要你做的:无,席位落地。

What changes

• ImportRowResultSchema (packages/spec/src/api/export.zod.ts) declares one optional member, errors: z.array(ValidateDataIssueSchema), one element per failed cell of a failed row. Its describe says the list holds the import's cell-coercion errors and that a cell error never carries the advisory-only rule or severity, that it is present only on a failed row with at least one cell error, that a row failing for another reason (a duplicate or ambiguous match key) omits it, and that the first-error keys field, code and error stay and name element 0. The three first-error describes now say they are element 0 when errors is present. The schema docblock gains one paragraph on the key.
• ImportResponseSchema.results and the async job's ImportJobResultsSchema.results carry it through the shared row schema; no other schema changes.
• Spec only: packages/core/src/utils/import-runner.ts is untouched. Filling the list is /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582's server half.
• Generated: packages/spec/authorable-surface/api.json gains api/ImportRowResult:errors; content/docs/references/api/export.mdx regenerated by gen:docs. A minor changeset for @objectstack/spec (Metadata-protocol: added: the authorable surface gains api/ImportRowResult:errors).

Why

Ruling 6377844879 on #25582 (batch #9 item 3a, letter A), quoted as written: 「一行所有单元格错误一次报全（报告加 errors: [{ field, code, message }]，原有第一错字段保留）」. Without a declared key, ImportRowResultSchema.parse strips one, so every reader typed by the schema would lose the list.

Element shape: ValidateDataIssueSchema, reused

Measured against what the issue asked (reuse unless there is a reason not to):

• Its required members are exactly the ruling's { field, code, message }, all z.string().
• The producer's element fits it as is: coerceRow returns FieldCoerceError[] (packages/core/src/utils/import-coerce.ts), { field: string; code: FieldErrorCode; message: string }, and FieldErrorCode is a string enum, so the server half can copy the list without mapping.
• The same row already uses it for warnings, and the validate verdict's row errors (ValidateDataResponseSchema, protocol.zod.ts) is ValidateDataIssueSchema[] too, so a row speaks one issue shape and a dry-run reader needs no third type.
• The only cost: the element also declares the advisory-only optional rule and severity. A cell error never carries them (the docblock says so), and both are optional, so nothing is required that a cell error cannot give. Not a reason to mint a new shape.

The test pins the element by identity (toBe(ValidateDataIssueSchema), and equal to the warnings element) and by type (expectTypeOf against ValidateDataIssue).

Tests

New describe('ImportRowResultSchema — errors (every cell error of a failed row)') in packages/spec/src/api/export.test.ts, 7 tests: parse keeps every element in order; field / code / error agree with element 0 after parse; the key is optional (a first-error-only failed row parses unchanged, no errors key); the element is ValidateDataIssueSchema itself; an element without message is refused at ['errors', 0, 'message']; ImportResponseSchema and ImportJobResultsSchema both keep it; the typed reader sees ValidateDataIssue.

Acceptance notes

• For /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582's server half (carrier: /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582). On a dry-run row the engine's validate verdict rejects, the runner sets error to composeValidationMessage(verdict.errors) (import-runner.ts, about :1293), every finding joined by ; , not element 0's message. A row filled from coerceRow's list (:1221) keeps error equal to element 0's message. The describe now scopes the list to the cell-coercion errors, so /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582 fills it from coerceRow only (option A, confirmed by the spec seat); filling it from verdict rows later would be a visible spec change.
• No refine enforces "present only on a failed row" or the element-0 agreement: a refine would refuse rows that parse today, which is a narrowing. The describe states the contract, the test pins it on a conforming row.
• check:dev-prereqs (local-only family) is red on this container because 37 workspace packages (mostly ui/) are not built; it judges local workspace state, not this diff. packages/spec/dist itself is fresh (the declaration stamp matches, per check:generated).

Verification

Local readings below were taken at head 2c2ade72; the respin at eff10df5 (changeset grade and describe sentence only) re-ran the build, gen:docs, check:generated, the changeset gates and the spec suite, recorded in the next block.

spec test, typecheck, generated artifacts

• pnpm --filter @objectstack/spec test (under the verify lock): Test Files 727 passed (727), Tests 22016 passed | 1 todo (22017), lock VERDICT command-exit 0.
• pnpm --filter @objectstack/spec typecheck: VERDICT command-exit 0; check:test-typecheck: OK, debt ledger held.
• pnpm --filter @objectstack/spec check:generated: first run named content/docs/references/** stale; after gen:docs, ✓ All 14 generated artifacts are up to date.

respin at eff10df (Metadata-protocol added, minor, describe sentence)

• pnpm --filter @objectstack/spec build && gen:docs (lock): VERDICT command-exit 0; check:generated: ✓ All 14 generated artifacts are up to date.
• pnpm --filter @objectstack/spec test (lock): Test Files 727 passed (727), Tests 22016 passed | 1 todo (22017), VERDICT command-exit 0.
• node scripts/check-changeset-no-major.mjs --base origin/main with this body as the event: exit 0, LEVEL AXIS: this PR declares clause-② yes (widening) and metadata-protocol added, and no package ... is graded patch. check-adr-0087-registration: exit 0, no declared-breaking changeset.

ablation (one, on the tested behaviour)

node scripts/ablation-replace.mjs renamed the declared key (errors: to ablatedErrors:, anchor 1 to 0, blob ee8a1e9b6078 to 69e10a948a46) and ran vitest run src/api/export.test.ts: Tests 5 failed | 52 passed (57); the five red were the parse, element-0 agreement, element identity, refusal and response/job tests. The optional-key test stayed green, as it should. Restored: blob equals HEAD and git diff HEAD empty. The test imports ./export.zod from source, so no dist leg applies.

derived gate families (dispatch-gates)

node scripts/pm/dispatch-gates.mjs --commands derived 106 commands from 5 changed paths (142 changed lines). All 106 ran; reconciliation with exit codes recorded: 106 derived famil(ies) accounted for — 99 run, 7 NOT-MEASURED (7 DERIVED from a recorded exit 3).

NOT MEASURED (exit 3, a prerequisite the gate refused on, none about packages/spec):

• pnpm --filter @objectstack/lint run check:doc-formula-expressions, check:doc-security-posture: @objectstack/formula / @objectstack/lint not built.
• pnpm --filter @objectstack/spec run check:skill-examples: packages/client-react/dist not built.
• pnpm check:docs-transcript-drift: @objectstack/lint not built.
• pnpm check:dual-build-cjs-loads, pnpm check:lean-entry-closure: other packages' dist/ not built.
• node scripts/check-plugin-teardown-shape.mjs --self-test: shallow checkout (checker health only).

CI builds the workspace fresh and runs these families; their verdicts are the PR's check-runs.

Human Minimum

• 1. Schemas start as Zod; packages/spec holds no runtime logic.
• 2. Every name I added uses the casing for its kind.
• 3. No formatter ran; pnpm lint is CI's run (repo-wide scan, not run locally).
• 4. One issue, one branch off an up-to-date main, one pull request.
• 5. A change that publishes carries a changeset, and both declaration lines are filled in here and in the changeset.
• 6. packages/spec touched: its generated artifacts are regenerated and committed.
• 7. No hand edits to content/docs/references/, packages/*/CHANGELOG.md or content/docs/releases/.
• 8. Title, body and comments are in English (the 维护者速读 block is Chinese by fleet convention).

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

1 anchor(s) derived from 1 changed package(s); no hand-written page names any of them. ⚠️ 1 changed file(s) yielded no anchor (packages/spec/authorable-surface/api.json), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files.

What this run could not see

• 1 changed file(s) yielded no anchor (packages/spec/authorable-surface/api.json) — pages documenting those are invisible to this run
• the SDK route bridge reached 55 of 207 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 — 146 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 775a0006bddbf7a1afc88b3a83e52cc0eea96711 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 72b207b93a8f3ce0af5e434ef441903042a24314 — the merge of head eff10df553173382ac5f2164b152375803ad4214 into base 775a0006bddbf7a1afc88b3a83e52cc0eea96711, 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 72b207b93a8f3ce0af5e434ef441903042a24314 && git checkout 72b207b93a8f3ce0af5e434ef441903042a24314
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 775a0006bddbf7a1afc88b3a83e52cc0eea96711 eff10df553173382ac5f2164b152375803ad4214 && git checkout -B drift-repro 775a0006bddbf7a1afc88b3a83e52cc0eea96711 && git merge --no-ff eff10df553173382ac5f2164b152375803ad4214

node scripts/docs-audit/affected-docs.mjs --json 775a0006bddbf7a1afc88b3a83e52cc0eea96711

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

[objectstack-fleet]

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 2c2ade72c8b717a8a51d357648570e4b7224660e
Local-runs: none
Scope: packages/spec/src/api/export.zod.ts + packages/spec/src/api/protocol.zod.ts, packages/core/src/utils/import-runner.ts, packages/spec/authorable-surface/api.json, scripts/check-changeset-no-major.mjs, ui/packages/plugin-grid/src/ImportWizard.tsx

① Derived judgments

1. Accept set: widening only, named right. ImportRowResultSchema is a plain z.object with no .strict(), so before this head an errors key was stripped and after it the key is kept. Nothing is refused that parsed before: the only new refusal is an errors element that is not { field, code, message }, and no server sends errors today. ImportResponseSchema.results and ImportJobResultsSchema.results carry it through the shared row schema. Clause-②: yes (widening) is the right arm. Correct.

2. Consumers typed by the row: none breaks. packages/core/src/utils/import-runner.ts:79 keeps its own ImportRowResult interface, not derived from the schema, so core compiles unchanged; REST serves core's rows and has no parse path through the schema (packages/rest/src/import-row-schema-drift-20701.test.ts reads keys by name). packages/client/src/index.ts:7136 types the answer as ImportResponse and only unwraps it. ui/packages/plugin-grid/src/ImportWizard.tsx:558 reads only the row's field, error and code (a Pick of those three) and :908 dedupes on result.field; both still read element 0's answer. plugin-auth IdentityImportRowResult extends ImportRowResult and app-shell intersect it; an optional key added to the base is additive for both. type-alias-convention.pin.test.ts:509 holds because ValidateDataIssueSchema has no default or transform (input equals infer), which warnings already proved. Correct.

3. Element shape: reusing ValidateDataIssueSchema is right. Its required members are exactly the ruling's { field, code, message }; the same row's warnings and the validate verdict's row errors (protocol.zod.ts:2607) already use it for errors, so there is precedent for an error list carrying the advisory-only rule and severity as optional. The cost is small and real: the generated nested-shape table in export.mdx now says rule and severity are "present only on an advisory rule hit" under ImportRowResult.errors[number], and nothing in the describe says a cell error is never such a hit. The docblock says it, but docblocks do not reach gen:docs. One sentence in the errors describe ("an element never carries rule or severity") would make the docs honest. Minor, not a FAIL.

4. The element-0 invariant is stated in a way the runner can keep only on one of its failed-row paths. At origin/main the runner fills a failed row from four sources: coerceRow cell errors (:1221, where error is errors[0].message, so the invariant holds); a duplicate or ambiguous match key (no cells, so the describe's "omits it" clause covers it); a rejecting validate verdict (:1293, where error is composeValidationMessage(verdict.errors), every finding joined by "; ", field is omitted when the first finding has none, and code falls back to VALIDATION_FAILED); and a refused write via toFailedResult / adoptDoorVerdict (error is the door's sentence). On the last two, filling errors would break "error names element 0". The describe says "present only on a failed row that has at least one cell error" but never says which list fills it. The ruling's scope is cell errors and the card names coerceRow as the producer, so the contract is keepable under option A. Finding: the describe should name the source (the coercion's cell errors), so a verdict-row fill is visibly a spec change and not a quiet drift. Carried with the respin below, not a FAIL on its own.

5. Docs: content/docs/references/api/export.mdx at head matches the zod describes in all three row tables (ImportJobResults.results, ImportResponse.results, ImportRowResult) and adds the ImportRowResult.errors[number] nested shape in the right place. The check:generated --reconcile-only lane (lint.yml:5293) was still in progress at review time, so freshness is not confirmed by CI yet.

6. Tests: the seven new pins assert the key survives a parse (not bare success), element identity with ValidateDataIssueSchema and the warnings element, refusal at ['errors', 0, 'message'], and the type. The element-0 test pins agreement on a conforming row only; no refine enforces it, which is right, since a refine would be a narrowing.

② Semver level

FAIL. The changeset grades @objectstack/spec patch with Metadata-protocol: none, and the PR moves packages/spec/authorable-surface/api.json (adds api/ImportRowResult:errors). scripts/check-changeset-no-major.mjs:2055 refuses exactly this: a declared none over a moved authorable-surface shard is verdict none-baseline-moved, exit 1, remedy "declare Metadata-protocol: added and grade the package at least minor; a baseline is never edited to quiet this gate". The head's Check Changeset check-run is red with that annotation (run 40449099225, job 122007327584). The dev report wrote the check as in_progress at report time; it has since failed.

Precedent agrees with the gate: the warnings widening of this same schema was graded minor (packages/spec/CHANGELOG.md, section at line 16392), and the last three commits that moved authorable-surface/api.json on main (b77bb37, f1ddace, f4112a7) all declared Metadata-protocol: added.

Byte-for-byte agreement: the PR body, the changeset and claim 6433227708 all read Clause-②: yes (widening) and Metadata-protocol: none. They agree with each other and are all wrong. The claim is the dispatch's own line; the dev copied it. Note the ruling's phrase "零元数据键" (no new authored key) is about authored metadata, and the ratchet still records api/ImportRowResult:* keys, so the gate is the arbiter here.

Respin needed: changeset '@objectstack/spec': minor, Metadata-protocol: added, in the changeset, the PR body and a fresh claim line on the card.

③ Boundary flags

• Open question (dev, verdict rows): answered in ① item 4. Option A for /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582's first cut is right and is the ruling's scope. Add to this PR's errors describe that the list is the coercion's cell errors, so the runner can keep the element-0 invariant and a later verdict fill is a declared spec change.
• Describe honesty on rule / severity: one sentence in the errors describe, as in ① item 3. Same respin.
• No refine for "present only on a failed row": agreed, a refine would narrow.
• check:dev-prereqs red locally: local-only family, judges workspace state, not this diff. Agreed.
• The dev's 7 NOT-MEASURED derived families all run in CI on this head: check:doc-formula-expressions (lint.yml:6472), check:doc-security-posture (lint.yml:6479), spec check:skill-examples (lint.yml:6465), check:docs-transcript-drift (lint.yml:5780), check:dual-build-cjs-loads (ci.yml:2941), check:lean-entry-closure (ci.yml:2965), check-plugin-teardown-shape --self-test (lint.yml:3931). Their lanes were in progress at review time.
• Check-runs at review: 42 on the head; 1 failure (Check Changeset), 8 Test Core shards, Build Core, Build UI Deps, Build Docs and 10 Lint and Type Check lanes still in progress; the rest success or skipped. Not all green.
• Author: PR user.login is objectstack-fleet[bot]; no non-fleet login to name.
• UI: the console still shows only element 0 (ImportWizard reads field / error / code). The business reason on /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582 is met at the wire only; a UI card to read errors is outside this card and the ruling.

Implemented-by: claude/project-thread-fb0xrv
Reviewed-by: session_01WGufzfhobWEgUUrjz4rrbh

Rendered by an isolated CONTRACT_REVIEW_TIER subagent of seat session_01WGufzfhobWEgUUrjz4rrbh (domain:spec#1), read-only, adversarial to the dispatch's own conclusions.

VERDICT: FAIL

[objectstack-fleet]

Contract review — addendum to record 6433766607 (FAIL at prior head 2c2ade7)

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: eff10df553173382ac5f2164b152375803ad4214
Local-runs: none
Scope: packages/spec/src/api/export.zod.ts + packages/spec/src/api/protocol.zod.ts, packages/core/src/utils/import-runner.ts, packages/spec/authorable-surface/api.json, scripts/check-changeset-no-major.mjs, ui/packages/plugin-grid/src/ImportWizard.tsx, .changeset/26425-import-row-every-cell-error.md, content/docs/references/api/export.mdx

This addendum judges the delta from the prior head 2c2ade7 to this head against the two findings of PR comment 6433766607; the ① judgments of that record stand unless an item below says otherwise.

① Derived judgments

1. Delta inventory: one commit, three files, 9 insertions and 7 deletions: the changeset (grade and Metadata-protocol line, plus one sentence in its prose), packages/spec/src/api/export.zod.ts (two string fragments inside the errors describe, no other token) and content/docs/references/api/export.mdx (the same sentence in the three errors rows). packages/spec/authorable-surface/api.json is unchanged in the delta, so the moved shard is still exactly api/ImportRowResult:errors; packages/spec/src/api/export.test.ts, protocol.zod.ts, import-runner.ts and ImportWizard.tsx are untouched. No schema member, type, default, refine or test moved, so prior items 1, 2 and 6 (accept set is widening only; no row-typed consumer breaks; the seven pins) hold unchanged. Correct.

2. Prior finding (2), describe honesty and source: met. The errors describe now reads, in the middle, "The list holds the import's cell-coercion errors, and a cell error never carries the advisory-only rule or severity." That names the producer (the coercion's cell errors, coerceRow at import-runner.ts:1221, where error equals errors[0].message) and states the rule/severity absence the docblock had kept to itself. With the source named, the element-0 invariant is keepable on the only path that fills the list, and a later fill from a verdict row (:1293, composeValidationMessage) or a refused write is a visible spec change, which is what prior item 4 asked for. The nested-shape table ImportRowResult.errors[number] in export.mdx still says rule and severity are "present only on an advisory rule hit": that is ValidateDataIssueSchema's own member describe, shared with warnings and the validate verdict, and the row-level sentence now says a cell error is never such a hit, so the two tables read consistently. Correct.

3. Generated docs match the describe: the describe string assembled from the four concatenated fragments at this head occurs verbatim in all three errors rows of content/docs/references/api/export.mdx at this head (the ImportJobResults.results, ImportResponse.results and ImportRowResult tables; compared by reading both files from the committed tree, no generator run). Prior item 5 is now confirmed on the committed tree rather than only claimed; the CI check:generated lane result is in ③.

4. Prior finding (1), byte-for-byte agreement of the declaration lines: met. Changeset at head: '@objectstack/spec': minor, Clause-②: yes (widening), Metadata-protocol: added. PR body (GitHub REST, pulls/26454, updated 04:36:22Z): Clause-②: yes (widening), Metadata-protocol: added, and its "What changes" prose says minor with Metadata-protocol: added. Claim amendment 6433776664 on spec: ImportRowResultSchema declares an optional errors array (every failed cell of a row), so /import can report more than the first error (#25582 proposal 4) #26425 (04:29:38Z): Metadata-protocol: added, Clause-②: yes (widening). Clause-② is unchanged on all three carriers, as required. Correct.

② Semver level

PASS. minor with Metadata-protocol: added is the grade scripts/check-changeset-no-major.mjs asks for over a moved authorable-surface shard, and it matches the warnings precedent on this same schema. The head's Check Changeset first ran at 04:33:13Z to 04:34:58Z against the pre-edit PR body and failed with the none-baseline-moved annotation (its metadata-protocol line read Metadata-protocol: none, the body's text at that minute, while the changeset already said added). The body was edited at 04:36:22Z and Check Changeset re-ran on this same head (job 122012337961, run 40450661111, 04:36:28Z to 04:38:08Z): success, no failure annotation. The earlier red job stays listed on the head's check-runs as a completed failure; the newest run of that check is green. Nothing in the delta changes the ① accept-set judgment, so the arm stays yes (widening).

③ Boundary flags

• Prior ③ answers stand: option A for /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582 is now written into the describe; no refine; check:dev-prereqs local-only; UI card outside this card.
• Dev open questions on the respin: none new. The body's "Acceptance notes" restate option A with the describe as the scope fence, which agrees with ① item 2.
• Check-runs on this head at 04:40:11Z: 49 listed. One completed failure, the superseded first Check Changeset job (122011379357); its re-run is success. Success since the prior record's reading: every Lint lane (Lint · CI, release & engine gates carries the teardown-shape self-test, lint.yml:3931), Type Check · source gates (carries check:generated --reconcile-only, lint.yml:5293, so the generated export.mdx is CI-fresh on this head), Build Docs, Plan Test Core shards, Spec property liveness, Governed Surface Queue Guard, Flag docs affected by code changes, Check Documentation Links and Dogfood Regression Gate. Still in progress: Build Core, Build UI Deps, Test Core shards 1 to 8, Type Check · consumer gates (check:doc-formula-expressions, lint.yml:6472), Type Check · debt ledger and Type Check · workspace (check:docs-transcript-drift, lint.yml:5780). Not all green yet; the landing waits on the in-progress lanes, not on this record.
• The prior record and this addendum both exist as comments naming different heads; the newest ## Contract review naming this head is this one.
• Author: PR user.login is objectstack-fleet[bot]; no non-fleet login to name.

Implemented-by: claude/project-thread-fb0xrv
Reviewed-by: session_01WGufzfhobWEgUUrjz4rrbh

Rendered by an isolated CONTRACT_REVIEW_TIER subagent of seat session_01WGufzfhobWEgUUrjz4rrbh (domain:spec#1), read-only, adversarial to the dispatch's own conclusions.

VERDICT: PASS

[objectstack-fleet]

席位核对 · ACCEPT · domain:spec seat 1 (#6017) · session session_01WGufzfhobWEgUUrjz4rrbh · 2026-10-27T05:10Z

Head eff10df553173382ac5f2164b152375803ad4214. Card #26425 (ruling 6377844879 on #25582, letter A), claim 6433227708 with amendment 6433776664.

• Contract review: owed at CONTRACT_REVIEW_TIER (an authorable key enters @objectstack/spec). FAIL 6433766607 at 2c2ade72 found the declaration wrong: Metadata-protocol: none over a moved authorable-surface/api.json, which was the seat's claim error. Addendum PASS 6433920008 names this head. It finds the changeset, the PR body and the amended claim agreeing on Clause-②: yes (widening) / Metadata-protocol: added with '@objectstack/spec': minor. It also finds that the errors describe names its source (the import's cell-coercion errors, with no rule/severity) and that this sentence appears in all three generated export.mdx rows.
• Shape: base main, Fixes #26425, 5 files. The change is spec only, and import-runner.ts is untouched. check-governed-merges --pr 26454: NOT governed, 144 changed lines.
• Gates: check-runs on the head are 42 success and 9 skipped. Check Changeset has two runs on this head. Run 122011379357 failed at 04:34 against the pre-edit body. The re-run 122012337961 after the 04:36 body edit is success. No other lane is red. mergeable_state is clean.
• Carried to /import: common spreadsheet cell shapes fail whole rows (dotted / compact dates, option value case, one error per row) #25582 (server half): the runner fills errors only from coerceRow's cell errors (option A), where error already equals element 0's message. A fill from verdict rows or refused writes would be a spec change under this describe.
• Landing: Tier S, not governed. pr_ready + automerge_enable in this act.