[objectstack-fleet]

Requested by Warren · project thread
Fixes #29335

Clause-②: no
Metadata-protocol: none

维护者速读

• 改了什么：只改 packages/spec/src/stack.zod.ts 里 TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL 的注释，外加一个 patch changeset。
• 为什么改：注释说顶层 datasets 由 AppPlugin 直接从 bundle 读取，树上没有这个读者；包体的 datasets 实际由引擎 registerApp 的 METADATA_ARRAY_KEYS 循环读取。
• 风险与回滚：零运行时风险，没有代码、类型或导出变化；回滚即 revert 本 PR。
• 先看：packages/spec/src/stack.zod.ts 的那段注释（约 360 行）。
• 两分钟验证：git grep -n "No boot reader takes a top-level" packages/spec/src/stack.zod.ts，预期命中 1 行。
• 需要你做的：无，席位落地。

What changes

• The TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL docblock now names the readers the tree has. Seed data and jobs are read off the stack whole, top level merged with every body (resolveArtifactCollections, @objectstack/runtime): AppPlugin hands the seed data to the seed loader and schedules the jobs. datasets are read off the package bodies only, by the engine's registerApp (METADATA_ARRAY_KEYS, @objectstack/objectql), and no boot reader takes a top-level datasets item that no body declares.
• The constant, the rule it encodes and every caller are unchanged.
• .changeset/29335-residual-datasets-docblock.md: @objectstack/spec patch. The package ships src/**/*.zod.ts as source (files[]), so the comment publishes even though dist/ carries none of it (measured: the new sentence has 0 hits in dist/; the constant's name, the positive control, has hits in dist/).

Readers measured on the tree at 94d3e254 (the branch's base before the merge of origin/main):

• packages/runtime/src/app-plugin.ts: no read of datasets off the bundle or its collections. Each datasets word there is the seed data collector (seedDatasets, readSeedDatasets, mergeSeedDatasets).
• packages/objectql/src/engine.ts: METADATA_ARRAY_KEYS lists datasets; registerApp runs packageBodyCollectionRefusal and then registerMetadataCollections over the body's keys.
• packages/metadata/src/plugin.ts: ARTIFACT_FIELD_TO_TYPE has no datasets row, so unclaimedTopLevel never lists one.

Tests

• No test changed: the diff is a comment. pnpm --filter @objectstack/metadata exec vitest run --maxWorkers=2 src/owner-stamp-residual-build-boot.test.ts (the test that pins the constant equal to the boot's rule): 1 file, 27 tests passed, at head 061f69c58.
• pnpm --filter @objectstack/spec typecheck: exit 0, at head 061f69c58.

Acceptance notes

• The boot-side half of the card (a packages[] stack's top-level datasets item that no body declares) is not fixed here, per triage's scope. It was measured on a kernel boot and is handed to the spec seat to file as a domain:server card: the item is dropped with no refusal and no warning.
• The same wrong attribution sits in a gate waiver: scripts/check-stack-collection-maps.mjs, the ARTIFACT_FIELD_TO_TYPE row for datasets, jobs, datasources and translations, says all four are consumed functionally by AppPlugin straight off the bundle. For datasets that is not so. Outside this card's file surface; reported to the seat, not edited.

Verification

node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack on this diff derived 74 families at 061f69c5. All 74 were run; 73 exit 0. pnpm check:dual-build-cjs-loads exits 3 (PREREQUISITE NOT MET: it needs every workspace package built), so it is NOT MEASURED locally and left to CI. --ran reconciliation: "74 derived famil(ies) accounted for — 73 run, 1 NOT-MEASURED (1 DERIVED from a recorded exit 3)".

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 passes. (No formatter ran. pnpm lint is the repo-wide scan CI owns; 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 Before Submitting list in CONTRIBUTING.md is done.

🤖 Generated with Claude Code

https://claude.ai/code/session_01JQwxbad6xiDCcLYXcBsEa6

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

⚠️ 1 changed file(s) yielded no anchor (packages/spec/src/stack.zod.ts), so the pages documenting them are NOT COVERED by this run — this is not a clean bill of health for those files. Nothing else in this diff resolved to a documentable surface (no symbol, route or SDK anchor derived from 1 changed package(s)).

What this run could not see

• 1 changed file(s) yielded no anchor (packages/spec/src/stack.zod.ts) — pages documenting those are invisible to this run
• 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 — 168 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 c88467d4adc417e668474e1b6198a5a7c503ac3f → packageMentionDocs.

[objectstack-fleet]

domain:spec#3 · os-warren · session_01JQwxbahe4v7GBD94QeSnir

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 061f69c580626603cdcd94e10d649049a270bdc5
Local-runs: none
Scope: packages/spec/src/stack.zod.ts + packages/metadata/src/plugin.ts, packages/runtime/src/app-plugin.ts, packages/runtime/src/artifact-collections.ts, packages/objectql/src/engine.ts, packages/objectql/src/plugin.ts, packages/metadata/src/owner-stamp-residual-build-boot.test.ts, scripts/check-stack-collection-maps.mjs

Inputs read: card #29335 (body, triage 6577799350, claim 6582652162, dev report 6583541780), PR #29514 (body, 2 files, net diff against main at the head), the 41 check-runs on the head. The head's merge-base with main is 86d36e99; no main commit since touches stack.zod.ts, and the three-dot diff is the two files alone.

① Derived judgments

• Accept set: unchanged — right. The only code line in the hunk's neighbourhood, TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL = new Set(['data', 'datasets', 'jobs']) (L373 on main), is not touched; neither are isResidualTopLevelItem, residualSlotOf or applyCollectionItemOwnerStampGate. ObjectStackDefinitionSchema accepts and refuses exactly what it did.
• Public surface: unchanged — right. The constant is module-private (no export), so no index.d.ts, json-schema, liveness ledger or api-surface line moves; Spec property liveness concluded success and no regenerated artifact was owed.
• Each claim the new comment makes, read in the cited code — right:
  • ARTIFACT_FIELD_TO_TYPE (packages/metadata/src/plugin.ts L130-175) has no data, datasets or jobs row, and unclaimedTopLevel (L415-424) walks exactly that map, so no top-level item of the three is ever listed as residual.
  • Seed data: AppPlugin reads this.collections.data (app-plugin.ts L1720-1745), where collections is resolveArtifactCollections(this.bundle) (L307); that resolver (artifact-collections.ts L457-490) merges the top-level value with every body's over packageOwnedCollectionKeys(), the keys shared by ObjectStackDefinitionSchema and AssembledPackageBodySchema, and data, jobs, datasets are all concat body collections (stack.zod.ts L1617-1641). "Read off the stack whole, the top level merged with every body" is exact.
  • jobs: collectBundleJobs reads resolveArtifactCollections(bundle).jobs (L2681) and AppPlugin arms scheduleAppArtifactJobs on kernel:ready (L1675-1705). "Schedules the jobs" is exact.
  • datasets: METADATA_ARRAY_KEYS holds 'datasets' (engine.ts L4444); registerApp runs packageBodyCollectionRefusal (L9295) and then registerMetadataCollections (L9481, loop L9594) over that list. The manifest service hands registerApp the ordered package bodies only (objectql/src/plugin.ts L84-85, L945) and takes the residual through unclaimedTopLevelObjects / unclaimedTopLevelItems, both ARTIFACT_FIELD_TO_TYPE-walked. "Read off the package BODIES only, by the engine's registerApp" is exact.
  • "No boot reader takes a top-level datasets item that no body declares": app-plugin.ts reads no datasets collection (every datasets word there is the seed-data collector, L80, L1720-2102, seed-datasets.ts); a grep of packages/{runtime,metadata,objectql,kernel,server}/src finds no other reader. The resolver does merge a top-level datasets item into collections.datasets, but nothing reads that key, so the sentence holds.
  • "A package BODY's items of these collections are still judged, at both doors" (kept from the old text): build door judgeBody over OWNER_STAMP_JUDGED_COLLECTIONS per body (stack.zod.ts L329-357); boot door packageBodyCollectionRefusal over METADATA_ARRAY_KEYS (engine.ts L5060). Still right.
• What the old text got wrong is gone — right. "datasets consumed by AppPlugin straight off the bundle" named a reader that does not exist on the tree; the card's one in-scope finding.

② Semver level

• .changeset/29335-residual-datasets-docblock.md: @objectstack/spec: patch — right. packages/spec/package.json files[] ships src/**/*.zod.ts, so the comment publishes in the tarball even though dist/ carries none of it; AGENTS.md Post-Task step 3 reserves skip-changeset for a diff that publishes nothing from any released package, and Metadata-protocol: none takes patch.
• Clause-②: no on the PR body and in the changeset, no arm; Metadata-protocol: none on both — right: no schema, type, export or authorable key moves. Check Changeset concluded success.

③ Boundary flags

• open_questions: [] — none to answer.
• Dev flag 1 (boot-side silent drop of a packages[] top-level datasets item no body declares): measured on a kernel boot and handed to the seat to file as domain:server, which is exactly what triage 6577799350 ruled ("not this card's"). The card's second Done when bullet is therefore deliberately open on this PR by that ruling; the Fixes #29335 close is consistent with it. Escalated to the seat — right.
• Dev flag 2 (scripts/check-stack-collection-maps.mjs L664-667, the ARTIFACT_FIELD_TO_TYPE waiver row, says datasets is "consumed functionally by AppPlugin straight off the bundle"): confirmed on main; it is the last restatement of the wrong attribution outside this PR. Outside the claimed file surface, reported not edited. Escalated to the seat — right; see ④ for a sibling.
• Human Minimum item 3 unchecked (pnpm lint not run locally): CI's Lint family owns it and was in_progress at read time; comment-only diff, no formatter ran.
• Docs Drift Check: no anchor for stack.zod.ts; nothing in content/docs/ documents a module-private constant's comment. No action.
• Check-runs on the head at read time: 41 total; every concluded run is success (Auto Label, Dogfood Regression Gate, Plan Test Core shards, Lint · PM dispatch-gates, Lint · scoped · tooling self-tests, filter, Spec property liveness, Check Changeset, PR-body guards, Flag docs affected, Check PR Size, Governed Surface Queue Guard) or skipped (opt-in and path-gated jobs); 18 still in_progress (Test Core 1-8, Build Core, Build UI Deps, Live E2E UI, the Lint and Type Check families). No gate has concluded against the head; the verdict below is judged on the concluded set and on a diff that changes no tested behaviour.

④ Design judgment

packages/spec/src/stack.zod.ts (the one hits file), judged against the scenario the card accepts it on — a comment that names the readers the tree has — and against every file it cites:

• Holds. Every reader the new comment names exists where it says, does what it says, and reads the shape it says (①). The sentence structure also now explains why the three are grouped although their readers differ (none is an ARTIFACT_FIELD_TO_TYPE item), which the old text blurred. The comment's pointer to the @objectstack/metadata equality test stays true: owner-stamp-residual-build-boot.test.ts L160-178 still asserts the not-residual set equals ['data', 'datasets', 'jobs'].
• Tried to break it: (a) collectBundleJobs is one-list-or-the-other against legacy manifest.jobs, not a merge — but the comment's "merged" is the top-level/body merge in resolveArtifactCollections, which is what it cites; not a defect. (b) A body's data has a second boot reader (registerApp step 6 registers seed data as metadata); the comment lists the reader that makes data never-residual and claims no exclusivity; not a defect. (c) The resolver does place a top-level datasets item into the merged collections.datasets; the comment says no reader takes it, and none does; not a defect.
• Should-fix (outside hits, carried to the seat, not blocking): packages/metadata/src/owner-stamp-residual-build-boot.test.ts L173-175, the comment on the notResidual assertion ("the three collections the boot reads off the stack whole, never as items"), restates the reading this PR just retired for datasets: the engine registers a body's datasets as items, and the top level is read by nobody. The assertion is right; only its gloss is stale, and this is the very test the new docblock points readers to. Folds into the same prose follow-up as the check-stack-collection-maps.mjs waiver row (③ flag 2). No must-fix found.

Remainder of the diff (the changeset file): the lane seat's ACCEPT at TIER_DEFAULT; judged here only as ②'s declaration, which is right.

Implemented-by: claude/project-thread-i4g3v9
Reviewed-by: session_01JQwxbahe4v7GBD94QeSnir

VERDICT: PASS

---

Generated by Claude Code

[objectstack-fleet]

domain:spec#3 · os-warren · session_01JQwxbahe4v7GBD94QeSnir

维护者速读

• 改了什么:只改 packages/spec/src/stack.zod.ts 中 TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL 的注释(+11 / -5),外加一个 @objectstack/spec patch changeset;常量、规则与所有调用方不变。
• 为什么改:旧注释说顶层 datasets 由 AppPlugin 直接从 bundle 读取,树上没有这个读者;包体里的 datasets 实际由引擎 registerApp 读取,顶层多出来的 datasets 没人读。
• 风险与回滚:零运行时风险,无代码、类型或导出变化;回滚即 revert。
• 先看:stack.zod.ts 约 361–375 行那段注释。
• 两分钟验证:git grep -n "No boot reader takes a top-level" packages/spec/src/stack.zod.ts,预期 1 行。
• 需要你做的:无,CI 全绿,席位落地。
• 席位意见:注释点名的每个读者都在树上核过(契约复核 6583699079 逐个给了行号);启动时静默丢弃顶层 datasets 的真缺陷已另立 boot: a packages[] stack's top-level datasets item that no package body declares is dropped with no refusal and no registration #29518,不在本 PR。

席位核对

ACCEPT · domain:spec#3 · head 061f69c580626603cdcd94e10d649049a270bdc5 · 2026-11-02T17:40Z

1. Shape: two files, packages/spec/src/stack.zod.ts (docblock only, +11 / -5) and .changeset/29335-residual-datasets-docblock.md (@objectstack/spec patch, both declaration lines). Inside the claim's file surface (claim 6582652162).
2. Premise, measured: on origin/main 94d3e254 the seat read the old sentence at stack.zod.ts ~L367 and found no datasets read in packages/runtime/src/app-plugin.ts beyond the seed collector. The contract review traced every reader the new text names on the live tree.
3. Gates: local 74 derived / 73 exit 0 / 1 NOT MEASURED (check:dual-build-cjs-loads, exit 3 on a prerequisite, left to CI). On this head every one of 64 check-runs is success or skipped; mergeable_state clean.
4. Accepted deviations: Fixes / Clause-② / Metadata-protocol sit above ## 维护者速读 (os-dev order). Human Minimum item 3 unchecked: pnpm lint is CI's, green here.
5. Out of scope: the boot-side silent drop of a top-level datasets item → filed boot: a packages[] stack's top-level datasets item that no package body declares is dropped with no refusal and no registration #29518 (finding, for triage to route; the dev's measurement is in report 6583541780). The same wrong attribution in the scripts/check-stack-collection-maps.mjs waiver row and the review's should-fix on packages/metadata/src/owner-stamp-residual-build-boot.test.ts L173-175 ride boot: a packages[] stack's top-level datasets item that no package body declares is dropped with no refusal and no registration #29518 (rider 6583740976). Skills-docs: n/a (Metadata-protocol: none).
6. Landing: not governed (check-governed-merges --pr 29514: NOT governed, 26 changed lines), Clause-②: no, contract review at CONTRACT_REVIEW_TIER PASS on this head (6583699079). Seat lands it: pr_ready + automerge_enable.

---

Generated by Claude Code

[github-actions]

⛔ merge queue 构建失败 — 先分诊，再决定要不要重排

队列构建 41494474291 红了。队列跑的是全量套件（PR 侧 CI 只跑 affected 子集），
所以失败的测试可能在本 PR 没碰过的包里 —— 那不是重排能修的。每次盲目重排都会让排在后面的所有 PR 重建一轮。

分类：failure —— 按下面的日志分诊。

失败的 job（日志抽取，best effort）：

• Test Core (2/8) — 失败步骤: Run this shard's tests

  @objectstack/spec:test:      ✓ inverts the card's measured control: the three bogus keys now FAIL parse instead of adding zero warnings  307ms
    ↳ 失败原因: (这条 FAIL 之后 12 行内没有可识别的原因行 —— 点进 job 看)
  @objectstack/spec:test:  FAIL   local  scripts/dropped-refinements.test.ts > the committed ledger > names at least one site per entry, and its header totals match its body
    ↳ 失败原因: @objectstack/spec:test: AssertionError: expected 993 to be 998 // Object.is equality
  @objectstack/spec:test:repo:  FAIL   repo  scripts/strictness-ledger-doc.test.ts > the ledger and the generated counts agree > is checked in current — every shard on disk equals a fresh render, and no
    ↳ 失败原因: @objectstack/spec:test:repo: AssertionError: expected Map{ …(14) } to deeply equal Map{ …(14) }
  @objectstack/spec:test:repo:  FAIL   repo  scripts/strictness-ledger-doc.test.ts > the ledger and the generated counts agree > the read-time totals equal the sums of the shards' own rows
    ↳ 失败原因: @objectstack/spec:test:repo: AssertionError: expected 'untriaged: 1297 object site(s) across…' to be 'untriaged: 1296 object site(s) across…' // Object.is equality
  

↳ 失败原因 是判读的关键：超时（Test timed out in … / Hook timed out in …）多半是负载/时序，不是本 PR 的回归；
断言（AssertionError: …）才指向真实的行为改变。两者的 FAIL 行长得一模一样，只有这一行能区分。

⚠️ 断言这一侧有一类例外，判据是断言在测什么，不是它是不是 AssertionError。 断言的对象是产品行为（一个值、一个形状、一次拒收）⇒ 照上面读：真实的行为改变，去查，⛔ 不要重排掉；
断言的对象是这次实验自身的有效性前提（跑完的耗时、负载下的先后、任何只在时间预算内才成立的条件）⇒ 它跟超时是同一类，同样对负载敏感，重排一次是合法的判别手段。
识别是机械的：断言的消息或它比较的值本身点名了一段时长、一个时间戳、一个耗时计数。实测过的一对 —— AssertionError: SecurityPlugin.init() ran: expected false to be true 测的是产品行为（真回归）；
AssertionError: this run took over a second, so second-precision stamps could have differed too: expected 1006 to be less than 1000 测的是实验前提：它守护的那条不变式当时是绿的，同一个 head 原样重排一次即成功。
穿着 AssertionError 外衣的时间测量，仍然是时间测量。（⛔ 这只改「怎么读一次红」，不改「哪些测试可以重排」——后者由别处管。）

跨 PR 相同签名（24h，按失败测试文件聚合）：

• scripts/strictness-ledger-doc.test.ts — 24h 内已弹出 13 个 PR（feat(spec): declare the whole-deployment export, every installed package as installable packages in one read (PackageExportBundleSchema) #29100、feat(spec): push channel protocol face — device registration widening, stored device shape, device contract members, push route, APNs/FCM config by credentialRef #29218、fix(spec)!: refuse a dashboard globalFilters entry keyed dateRange beside a declared dateRange #29370、fix(spec)!: a datasource named drivers is refused at every door that parses a datasource #29423、feat(spec): a record page can name the apps whose routes open it (PageSchema.apps) #29437、docs(spec): tenancy-posture prose says which side of the open/commercial split gates activation #29454、feat(spec): declare features.sso on the auth config response #29455、build(turbo): hash the Service Worker generator into the Console's build and type-check only #29504、fix(pm): the relay reads no runner 403 as the egress proxy's, and a refused advisory row names its cause #29505、fix(cli): os migrate organization-ownership --apply carries a public-form withdrawal onto a read-only package view into its .custom pair, declared #29508、docs(spec): the disabledActions description names Setup › Packaged Actions, not the retired Packaged automation page #29511、docs(spec): TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL names the readers of top-level datasets the tree has #29514、feat(pm): a ruled decision card still in the box is a patrol row, and the director's ruling lands with its knock #29515），但它们同属 1 条投机栈 ⇒ 1 次独立命中。汇总 issue：Queue-flake anchor: scripts/strictness-ledger-doc.test.ts #29524（已刷新）
• scripts/dropped-refinements.test.ts — 24h 内已弹出 6 个 PR（feat(spec): a record page can name the apps whose routes open it (PageSchema.apps) #29437、docs(spec): tenancy-posture prose says which side of the open/commercial split gates activation #29454、fix(pm): the relay reads no runner 403 as the egress proxy's, and a refused advisory row names its cause #29505、docs(spec): the disabledActions description names Setup › Packaged Actions, not the retired Packaged automation page #29511、docs(spec): TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL names the readers of top-level datasets the tree has #29514、feat(pm): a ruled decision card still in the box is a patrol row, and the director's ruling lands with its knock #29515），但它们同属 1 条投机栈 ⇒ 1 次独立命中。汇总 issue：Queue-flake anchor: scripts/dropped-refinements.test.ts #29530（已刷新）
• ⚠️ 24h 评论账本没读完（超过 5 页仍未读到窗口尽头），所以上面的「不同 PR 数」是下界，不是全量。

历史信号：

• 本 PR 过去 24h 无队列失败记录（首次）。
• 过去 24h 队列共有 21 个失败构建（不含本次）。

分诊清单：

1. 失败测试在本 PR 改动的包里 → 真回归，修 PR。
2. 失败测试与本 PR 无关 → 看上面的「跨 PR 相同签名」；已有汇总 issue ⇒ flaky/环境问题实锤，去那张 issue 上谈，修好前重排只会再烧一轮全队列。
3. 两者都不是 → 可能与同组 PR 语义冲突；等前面的 PR 落地或失败出队后再重排一次即可，不要连续重排。

---

Generated by Claude Code · merge-queue-triage workflow (#4859)

[github-actions]

⛔ merge queue 构建失败 — 先分诊，再决定要不要重排

队列构建 41495594461 红了。队列跑的是全量套件（PR 侧 CI 只跑 affected 子集），
所以失败的测试可能在本 PR 没碰过的包里 —— 那不是重排能修的。每次盲目重排都会让排在后面的所有 PR 重建一轮。

分类：failure —— 按下面的日志分诊。

失败的 job（日志抽取，best effort）：

• Test Core (2/8) — 失败步骤: Run this shard's tests

  @objectstack/spec:test:  FAIL   local  scripts/dropped-refinements.test.ts > the committed ledger > names at least one site per entry, and its header totals match its body
    ↳ 失败原因: @objectstack/spec:test: AssertionError: expected 993 to be 998 // Object.is equality
  

↳ 失败原因 是判读的关键：超时（Test timed out in … / Hook timed out in …）多半是负载/时序，不是本 PR 的回归；
断言（AssertionError: …）才指向真实的行为改变。两者的 FAIL 行长得一模一样，只有这一行能区分。

⚠️ 断言这一侧有一类例外，判据是断言在测什么，不是它是不是 AssertionError。 断言的对象是产品行为（一个值、一个形状、一次拒收）⇒ 照上面读：真实的行为改变，去查，⛔ 不要重排掉；
断言的对象是这次实验自身的有效性前提（跑完的耗时、负载下的先后、任何只在时间预算内才成立的条件）⇒ 它跟超时是同一类，同样对负载敏感，重排一次是合法的判别手段。
识别是机械的：断言的消息或它比较的值本身点名了一段时长、一个时间戳、一个耗时计数。实测过的一对 —— AssertionError: SecurityPlugin.init() ran: expected false to be true 测的是产品行为（真回归）；
AssertionError: this run took over a second, so second-precision stamps could have differed too: expected 1006 to be less than 1000 测的是实验前提：它守护的那条不变式当时是绿的，同一个 head 原样重排一次即成功。
穿着 AssertionError 外衣的时间测量，仍然是时间测量。（⛔ 这只改「怎么读一次红」，不改「哪些测试可以重排」——后者由别处管。）

跨 PR 相同签名（24h，按失败测试文件聚合）：

• scripts/dropped-refinements.test.ts — 24h 内已弹出 7 个 PR（feat(spec): a record page can name the apps whose routes open it (PageSchema.apps) #29437、docs(spec): tenancy-posture prose says which side of the open/commercial split gates activation #29454、feat(lint)!: os lint refuses a flow region the engine refuses to register (flow-region-shape-invalid) #29495、fix(pm): the relay reads no runner 403 as the egress proxy's, and a refused advisory row names its cause #29505、docs(spec): the disabledActions description names Setup › Packaged Actions, not the retired Packaged automation page #29511、docs(spec): TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL names the readers of top-level datasets the tree has #29514、feat(pm): a ruled decision card still in the box is a patrol row, and the director's ruling lands with its knock #29515），但它们同属 1 条投机栈 ⇒ 1 次独立命中。汇总 issue：Queue-flake anchor: scripts/dropped-refinements.test.ts #29530（已刷新）
• ⚠️ 24h 评论账本没读完（超过 5 页仍未读到窗口尽头），所以上面的「不同 PR 数」是下界，不是全量。

历史信号：

• ⚠️ 本 PR 过去 24h 已在队列失败 1 次（不含本次）。 内容未变而反复失败 ⇒ 高度怀疑 flaky 测试或与同组 PR 的语义冲突，重排不解决。
• 过去 24h 队列共有 26 个失败构建（不含本次）。

分诊清单：

1. 失败测试在本 PR 改动的包里 → 真回归，修 PR。
2. 失败测试与本 PR 无关 → 看上面的「跨 PR 相同签名」；已有汇总 issue ⇒ flaky/环境问题实锤，去那张 issue 上谈，修好前重排只会再烧一轮全队列。
3. 两者都不是 → 可能与同组 PR 语义冲突；等前面的 PR 落地或失败出队后再重排一次即可，不要连续重排。

---

Generated by Claude Code · merge-queue-triage workflow (#4859)

[objectstack-fleet]

domain:spec#3 · os-warren · session_01JQwxbahe4v7GBD94QeSnir

Landed · 2026-11-02T18:46Z

Reading	Value
PR #29514	MERGED 2026-11-02 18:42:19Z via merge queue, merge commit 7eb0b19f on main
Card #29335	closed completed by Fixes #29335; no other closing keyword in the body

State change: pm:dispatched removed from #29335. Delivered: the TOP_LEVEL_COLLECTIONS_NEVER_RESIDUAL prose names the readers of top-level datasets the tree has. Remaining: the boot drop of a residual top-level datasets item is card #29518 (finding), not this card.

---

Generated by Claude Code