fix(service-analytics)!: refuse a caller-supplied analytics member that is not a column reference at the door

[objectstack-fleet]

Fixes #21177

Clause-②: no (narrowing)

What

The analytics doors evaluated caller-supplied content at query time that neither the object-level nor the field-level read admission could judge. After merging main, this PR is narrowed to the residual that main does not yet judge.

The /analytics/query half is superseded

main landed #21156 (as #21173): a caller-named member that is neither a declared cube member nor a column reference is refused at the /analytics/query door in every tier, through the single field-read judge (PERMISSION_DENIED / 403 — "one judge, one shape, no new error code"). This PR therefore:

• drops its own duplicate handling of that class, and
• reverts its edits to field-read-admission-gate.test.ts back to service-analytics: delete the analytics field gate's stand-down on an authored cube expression member once #20943 retires raw expressions in a cube member's sql (the branch becomes unreachable) #20965's pin as main has it.

The residual: a dataset's own field text

The remaining gap is content main does not judge: the dimension and measure field text of a dataset sent to POST /api/v1/analytics/dataset/query. The service compiles that dataset into a cube whose members read as declared, so a dimension's or measure's field that is a raw expression resolves to a declared cube member — and #21156 leaves a declared expression member to the field-level gate, which stands down with no security service and on an object its reader answers undefined for. In those tiers the expression reached the native statement as written.

This PR judges the dataset's own field text at the dataset door, before the dataset is compiled and before any strategy runs, through main's existing judge (assertCallerMembersJudgeable / fieldReadUnjudgeableError): PERMISSION_DENIED / 403, naming the member and never the expression text, for every caller (admin included) and whether or not a security service is wired. No new error code and no second admission module — caller-content-admission.ts is removed.

What is left to the existing gates, unchanged:

• The dataset's own filter, the selection's runtime filter and cube-query members are lowered into the compiled query and already judged on the query path by [security] An analytics query path can carry caller-supplied member text into the native statement in a tier the field gate does not judge — detail withheld pending maintainer #21156.
• A dataset registered through the configuration door and queried by cube name runs through query(), where [security] An analytics query path can carry caller-supplied member text into the native statement in a tier the field gate does not judge — detail withheld pending maintainer #21156 leaves its members to the existing gates.

Inline and saved alike. The route hands this door an inline dataset (body.dataset) and a saved one alike: it loads body.datasetName from metadata and calls the same queryDataset, and the build probe calls it with a saved dataset too. The service cannot tell the branches apart, so the refusal is uniform, which is the safer reading. A saved dataset whose field is an expression is refused the same way as an inline one. No shipped dataset carries a non-column field (examples/app-crm, examples/app-showcase, examples/app-todo and platform-objects were read; the showcase done_rate is a derived measure, not a field expression). Refusing such a field when it is authored belongs to the dataset schema's own retirement of expression fields, a separate change.

This aligns the caller-supplied dataset surface with ADR-0021's "zero raw expressions".

What stays answerable

Every dataset (inline or saved) and cube query that names columns, relationship paths and declared members — the whole legitimate author and query surface — is unchanged. A plain-column inline dataset, a saved plain-column dataset and a dataset registered through the configuration door are all still answered. Only a dataset field whose text is a raw expression is refused, inline or saved; the BREAKING note in the changeset covers both.

Scope / serial

• The fix is at the analytics door (analytics-service.ts), never in a strategy. read-scope-sql.ts, strategies/native-sql-strategy.ts and contains-membership-sql.ts (fix(service-analytics)!: the analytics read scope and the native where answer $contains on a JSON-stored field by membership (#20987) #21117 / [finding] $contains / $notContains on a declared multi-valued or JSON-stored field still answer SUBSTRING on five faces, the analytics RLS read scope among them (u1 admits a row storing u10) #20987) are untouched.
• origin/main was merged in (no rebase, no stacking). The residual gate sits beside [security] Family closeout: stored metadata bodies (stored datasource credential material included) are still reachable outside the redacting doors at several further positions — detail withheld pending maintainer #21120's stored-metadata-body refusal and [security] An analytics query path can carry caller-supplied member text into the native statement in a tier the field gate does not judge — detail withheld pending maintainer #21156's caller-member gate, on the same unconditional seam.

Tests

• inline-dataset-field-admission-door.test.ts — the residual, on the dataset door × both strategies × four provider tiers (no provider, admin-unrestricted, non-admin field list, reader answers undefined for the object): refused 403 PERMISSION_DENIED, the strategy/driver never called; a plain-column inline dataset and a dataset registered through the configuration door still answered.
• field-read-admission-gate.test.ts — reverted to main (service-analytics: delete the analytics field gate's stand-down on an authored cube expression member once #20943 retires raw expressions in a cube member's sql (the branch becomes unreachable) #20965's pin stays).
• packages/rest/src/analytics-16019-driver-declared-fault.test.ts — its first block drove a driver fault through exactly the inline-dataset field expression path this change closes, so it now pins the door refusal (403 PERMISSION_DENIED) plus a positive control. It also pins the route's saved branch end to end: a saved dataset reached through body.datasetName whose field is not a column reference is refused 403 PERMISSION_DENIED with the driver never called, and a saved plain-column dataset is answered 200. The looksLikeInternalErrorLeak recognises no such column: but not no such function: — a SQLite parse failure echoes the raw engine message into the 500 body #16019 relay stays covered by its second block and by driver-sql's own sql-driver-16019-raw-statement-fault-envelope.test.ts.
• Ablation: disabling the new seam on disk turns the residual pins red (the three tiers the field gate stands down in, and the saved-branch pin, which is then served 200) while the controls stay green; restored. The full @objectstack/service-analytics suite is green, as is the non-SQL temporal step under a skewed process zone.

Review

Security boundary. The maintainer has authorized landing; the first merge-queue build failed on the semantic overlap with #21156 described above, and this push carries the narrowing. Its draft/ready state is unchanged by this push, and it is not queued or armed for auto-merge here — the owning seat re-enqueues once CI is green on this head.

Thread receipts (round 5)

• Docs Drift Check (5934340689): computed on the previous head 5428f7e627; both release pages it lists were reached through a string literal in assertQueryMembersJudgeable, which this push deletes. Release pages are read-only here and none is edited.
• Landing note (5934790041): the maintainer's landing authorization on [security] An analytics query surface evaluates caller-supplied content outside the object- and field-level read admission, so a non-admin member can read data they are refused everywhere else — detail withheld pending maintainer #21177 is recorded there. The ready state and auto-merge it describes belong to the owning seat; this push does not change either.
• Merge-queue triage (5935112816, run 36885164397): triaged as case 1 — every failing test is in @objectstack/service-analytics, which this PR changes, and each is the semantic overlap with [security] An analytics query path can carry caller-supplied member text into the native statement in a tier the field gate does not judge — detail withheld pending maintainer #21156 (an expected INVALID_FIELD / 400 met main's PERMISSION_DENIED / 403). This push resolves both failing files: caller-content-admission-door.test.ts is removed with the duplicate /analytics/query handling, and field-read-admission-gate.test.ts is back to main's pin. The full @objectstack/service-analytics suite and the non-SQL temporal step (the two failing jobs' steps) pass locally on this head.

_{read 2026-10-01T17:23Z.}

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

This PR changes 1 package(s): @objectstack/service-analytics, touching 4 documentable anchor(s).

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

• content/docs/releases/v17/17-5.mdx (via AnalyticsService (symbol, a top-level class))

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

• 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 — 10 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 862f12c0b9b36a6192f6ffe55f4052fa7dbaf34a → packageMentionDocs.

Which tree this was computed on

This run read content/docs from c3663627264748d3580544ebba9518af85b1e6ea — the merge of head aea35ada9cbd0e1d9a642120f9a8b6ff519de6c4 into base 862f12c0b9b36a6192f6ffe55f4052fa7dbaf34a, 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 c3663627264748d3580544ebba9518af85b1e6ea && git checkout c3663627264748d3580544ebba9518af85b1e6ea
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 862f12c0b9b36a6192f6ffe55f4052fa7dbaf34a aea35ada9cbd0e1d9a642120f9a8b6ff519de6c4 && git checkout -B drift-repro 862f12c0b9b36a6192f6ffe55f4052fa7dbaf34a && git merge --no-ff aea35ada9cbd0e1d9a642120f9a8b6ff519de6c4

node scripts/docs-audit/affected-docs.mjs --json 862f12c0b9b36a6192f6ffe55f4052fa7dbaf34a

⚠️ 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 862f12c0b9b36a6192f6ffe55f4052fa7dbaf34a → pass the list as
args.docs, on the commit named under Which tree this was computed on.

[objectstack-fleet]

Landing note: this is a permission-boundary change (human floor), so the PR stayed draft until the maintainer decided. The maintainer authorized the landing on #21177, selecting verbatim 「绿了就转 ready + 进队列（推荐）」 in Claude Code session session_01MRdbfpy4sQT8bUjmMhxsN7 on 2026-10-01. They then gave a standing instruction for this session, verbatim: 「绿了就应该进队列合并啊，为什么一直问我」. CI on head 5428f7e627 is 31 success / 3 skipped / 0 failed. The PR is now ready with auto-merge armed, and the merge queue re-validates on the merged generation.

---

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 36885164397 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。

失败的 job（日志抽取，best effort）：

• Test Core (4/6) — 失败步骤: Run this shard's tests

  @objectstack/service-analytics:test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'NativeSQLStrategy' > 'no security pro
    ↳ 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/service-analytics:test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'NativeSQLStrategy' > 'admin — provide
    ↳ 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/service-analytics:test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'NativeSQLStrategy' > 'non-admin — pro
    ↳ 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/service-analytics:test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'ObjectQLStrategy' > 'no security prov
    ↳ 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/service-analytics:test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'ObjectQLStrategy' > 'admin — provider
    ↳ 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/service-analytics:test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'ObjectQLStrategy' > 'non-admin — prov
    ↳ 失败原因: @objectstack/service-analytics:test: AssertionError: expected Error: [Analytics] Access denied: member … { …(4) } to match object { code: 'INVALID_FIELD', …(3) }
  @objectstack/service-analytics:test:  FAIL  src/__tests__/field-read-admission-gate.test.ts > [#20965] the field-level read gate — a member that names no field is refused, never stood down > 'NativeSQ
    ↳ 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/service-analytics:test:  FAIL  src/__tests__/field-read-admission-gate.test.ts > [#20965] the field-level read gate — a member that names no field is refused, never stood down > 'ObjectQL
    ↳ 失败原因: @objectstack/service-analytics:test: AssertionError: expected Error: [Analytics] Access denied: member … { …(4) } to match object { code: 'INVALID_FIELD', …(2) }
  

• Temporal Conformance (live PG + MySQL) — 失败步骤: Run the non-SQL temporal backends under the skewed process zone

  packages/services/service-analytics test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'NativeSQLStrategy' > 'no securit
    ↳ 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)
  packages/services/service-analytics test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'NativeSQLStrategy' > 'admin — pr
    ↳ 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)
  packages/services/service-analytics test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'NativeSQLStrategy' > 'non-admin 
    ↳ 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)
  packages/services/service-analytics test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'ObjectQLStrategy' > 'no security
    ↳ 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)
  packages/services/service-analytics test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'ObjectQLStrategy' > 'admin — pro
    ↳ 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)
  packages/services/service-analytics test:  FAIL  src/__tests__/caller-content-admission-door.test.ts > [#21177] caller-content admission — the /analytics/query door > 'ObjectQLStrategy' > 'non-admin —
    ↳ 失败原因: packages/services/service-analytics test: AssertionError: expected Error: [Analytics] Access denied: member … { …(4) } to match object { code: 'INVALID_FIELD', …(3) }
  packages/services/service-analytics test:  FAIL  src/__tests__/field-read-admission-gate.test.ts > [#20965] the field-level read gate — a member that names no field is refused, never stood down > 'Nat
    ↳ 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)
  packages/services/service-analytics test:  FAIL  src/__tests__/field-read-admission-gate.test.ts > [#20965] the field-level read gate — a member that names no field is refused, never stood down > 'Obj
    ↳ 失败原因: packages/services/service-analytics test: AssertionError: expected Error: [Analytics] Access denied: member … { …(4) } to match object { code: 'INVALID_FIELD', …(2) }
  

↳ 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）：

• src/__tests__/caller-content-admission-door.test.ts — 24h 窗口内只有本 PR 撞到过，暂不汇总（再有一个不同 PR 撞到就会自动开汇总 issue）。
• src/__tests__/field-read-admission-gate.test.ts — 24h 窗口内只有本 PR 撞到过，暂不汇总（再有一个不同 PR 撞到就会自动开汇总 issue）。
• ⚠️ 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 1 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]

Re-landing note: the merge-queue run on the previous head failed because of a semantic overlap with #21173 (#21156) on main. That overlap is reconciled in rounds 5 and 6; the PM review record is on #21177. The PR stays ready. Auto-merge is re-armed on head aea35ada9c, so the PR enters the merge queue only after its required checks pass. This follows the maintainer's standing instruction for session session_01MRdbfpy4sQT8bUjmMhxsN7, verbatim: 「绿了就应该进队列合并啊，为什么一直问我」.

---

Generated by Claude Code