[objectstack-fleet]

Requested by Steve · project thread
Fixes #28803

Clause-②: no
Metadata-protocol: none

维护者速读

• 改了什么:fieldPatches 文档示例里的选项键从 tech 改成 crm_account.industry 真实存储的值 technology(technology: { label: 'Tech' }),覆盖 spec 描述、TSDoc、选项列表拒收文案、对象扩展指南和两份未发布的 changeset。
• 为什么改:原示例照抄会在 os validate 和安装时被拒(该字段没有 tech 这个选项),文档在教一个坏例子;技能包那份已在 docs(skills): objectstack-data teaches an object extension's fieldPatches — relabel a vendor field without redeclaring it in fields #28795 修好,这里补齐其余几份。
• 风险与回滚:纯文案,schema 接受/拒收的结果不变;回滚即 revert 本 PR。
• 先看:packages/spec/src/data/object.zod.ts(三处示例文字)、content/docs/data-modeling/object-extensions.mdx、content/docs/references/data/object.mdx(重新生成的参考页)。
• 两分钟验证:git grep -n "tech: {" -- packages/spec/src content/docs skills .changeset ':!*.test.ts',预期没有任何输出。
• 需要你做的:无,席位落地。

What changes

• packages/spec/src/data/object.zod.ts: the fieldPatches .describe() example, the ObjectExtensionSchema TSDoc example, and the refusal text for an option list under options now read options: { technology: { label: 'Tech' } }. crm_account.industry (examples/app-crm/src/objects/account.object.ts) stores technology, finance, healthcare, retail, other; tech is none of them, so a copy was refused with reason option.
• content/docs/data-modeling/object-extensions.mdx: the guide's example, same re-key.
• content/docs/references/data/object.mdx: re-recorded by check:generated --fix (only check:docs was stale).
• .changeset/28565-field-patch-applied.md and .changeset/28560-field-patch-slot.md: both still unreleased (present in .changeset/ on main at 1617b7f, no Version Packages release has consumed them), so their copies of the example are re-keyed too. The 28560 copy is a fifth copy the card did not list.
• .changeset/28803-field-patches-example.md: spec patch.

No forced test edits. Test fixtures that patch tech (field-customization.test.ts, field-patches-derived.test.ts, and the cli / metadata-core / objectql / cloud-connection field-patch tests) build their own objects and do not pin the documented example text, so they stay as they are.

Tests

• pnpm --filter @objectstack/spec exec vitest run --maxWorkers=2 src/data and pnpm --filter @objectstack/spec typecheck: Test Files 154 passed (154), Tests 4756 passed, 1 todo; VERDICT command-exit 0.
• pnpm --filter @objectstack/spec build && check:generated: 1 of 14 stale (check:docs), fixed with --fix, then all up to date.

Acceptance notes

• Prose only; no consumer package test is owed (no accept/reject result changes).
• Pending release notes corrected, so Check Changeset stays red by design. This PR edits two changesets it did not add, both still pending (no Version Packages release has consumed them): .changeset/28565-field-patch-applied.md (named by the card, its example line) and .changeset/28560-field-patch-slot.md (a fifth copy of the same example, not named by the card). In each, only the option key and label inside the code example change, from tech: { label: 'Technology' } to technology: { label: 'Tech' }; no sentence about behaviour changes. This is the DELIBERATE CORRECTION class check-empty-changeset.mjs names: the gate stays red, skip-changeset is not applied, and the correction needs a written confirmation on this PR. Restoring either file from the base would republish an example that os validate refuses.
• Merged origin/main at 663b7c0 with scripts/pm/os-regen-merge.sh (fix(objectql,cloud-connection)!: a field patch reaches the owner's own fields only, at the fold and the install door; the install door refuses it with 422 (#28733) #28791 had edited object.zod.ts and object-extensions.mdx); no conflict, and check:generated reports all 14 artifacts up to date on the merged head.

Verification

All readings on head 0c29908fd8 (the merge of main).

spec tests and typecheck

bash scripts/pm/os-verify-lock.sh -c 'pnpm --filter @objectstack/spec test && pnpm --filter @objectstack/spec typecheck': Test Files 788 passed (788), Tests 23268 passed, 1 todo; VERDICT command-exit 0.

generated artifacts

Before the merge: check:generated found 1 of 14 stale (check:docs, content/docs/references/data/object.mdx), fixed with --fix. After the merge: pnpm --filter @objectstack/spec build && pnpm --filter @objectstack/spec check:generated: all 14 generated artifacts up to date; VERDICT command-exit 0.

derived gate list

node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack listed 97 commands; each ran on this head with its exit code recorded. 96 exit 0. node scripts/check-empty-changeset.mjs --base origin/main exits 1 on the two pending changesets above, as designed. Reconciliation: 97 derived, 97 run, 0 NOT-MEASURED, 0 UNRUN (a derived zero, every line carries an exit code). Seven gates first exited 3 (prerequisites: unbuilt packages, shallow clone); after a full turbo run build and git fetch --unshallow all seven exit 0. Local-only: check-dev-prereqs exit 0, check:skill-frame-freshness exit 0.

---

Generated by Claude Code

[github-actions]

📓 Docs Drift Check

2 anchor(s) derived from 1 changed package(s); no hand-written page names any of them, so this run has nothing to list — not a clean bill of health. This check sees only pages that NAME a derived anchor: one that documents this change in prose, or enumerates it in an authoring dialect, names none and stays invisible to it on every run.

What this run could not see

• the SDK route bridge reached 60 of 216 client-bound route-ledger rows — the other 156 have no registrar path: tail to select them, so pages documenting THEIR client methods cannot appear above, on this or any run. Of those 156: 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; 101 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 — 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 3e05ed1ec2ecce09b19a23bd51a6f14afb40caf3 → packageMentionDocs.

Which tree this was computed on

This run read content/docs from 38eb084b7568068953b60667259572602449ebbf — the merge of head 0b024ae90c2013c870e437ff844dd15af785c688 into base 3e05ed1ec2ecce09b19a23bd51a6f14afb40caf3, 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 38eb084b7568068953b60667259572602449ebbf && git checkout 38eb084b7568068953b60667259572602449ebbf
# afterwards, rebuild it from the two parents, which stay fetchable
git fetch origin 3e05ed1ec2ecce09b19a23bd51a6f14afb40caf3 0b024ae90c2013c870e437ff844dd15af785c688 && git checkout -B drift-repro 3e05ed1ec2ecce09b19a23bd51a6f14afb40caf3 && git merge --no-ff 0b024ae90c2013c870e437ff844dd15af785c688

node scripts/docs-audit/affected-docs.mjs --json 3e05ed1ec2ecce09b19a23bd51a6f14afb40caf3

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

[objectstack-fleet]

domain:spec#1 · os-steve · session_01ADHkmPi25XXKx92Xmg4viS

Head-sha: 0b024ae90c2013c870e437ff844dd15af785c688 · read 2026-11-01T22:15Z

维护者速读

• 改了什么:fieldPatches 文档示例的选项键由 tech 改为 crm_account.industry 真实存储的 technology,共 5 处：spec 描述、TSDoc、选项列表拒收文案、对象扩展指南，以及两份尚未发布的 changeset(28560、28565)。参考页已重新生成。
• 为什么改：旧示例照抄会在 os validate 和安装时被拒，因为该字段没有 tech 这个选项。
• 风险与回滚：纯文案，接受与拒收结果都不变。回滚即撤销本 PR。
• 更正 PR 正文速读第 2 行：技能包那份示例在 docs(skills): objectstack-data teaches an object extension's fieldPatches — relabel a vendor field without redeclaring it in fields #28795 里改，该 PR 尚未合并,不是「已修好」。新 changeset 里原先「与已发布技能同拼写」一句，已应席位要求删除(head 0b024ae9)。
• 需要你做的：无。PR 正文要你书面确认对两份待发 changeset 的更正。按落地规则，这由同 head 的契约复审 PASS 记录确认，不用你动手。这个文件在 packages/spec/src 下，已交分诊复审;PASS 且 CI 全绿后由席位落地。

席位意见：可落地，等契约复审。

席位核对 — ACCEPT · domain:spec#1 · head 0b024ae9

1. Shape: draft, base main, first line Fixes #28803, no other closing keyword. 7 files, all inside the claimed surface: object.zod.ts (3 prose strings), object-extensions.mdx, the regenerated references/data/object.mdx, the new spec patch changeset, and the two pending changesets. Not governed. Prose only: the three object.zod.ts hunks change a .describe() string, a TSDoc example and an error-message string. No schema shape, key or refinement moves.
2. Premise, measured on origin/main: examples/app-crm/src/objects/account.object.ts lists industry values technology, finance, healthcare, retail, other; tech is not one of them. On the head, git grep "tech: {" over packages/spec/src content/docs skills .changeset (tests excluded) hits nothing except the new changeset's quote of the old text.
3. Gate check-runs (pinned to head 0b024ae9): at reading 7 success, 3 skipped, 13 running, 0 failed. Check Changeset is expected red (④ ①). Landing needs every other check green or an expected skip. Dev readings: on 0c29908f, spec tests 788/788 files and typecheck exit 0; check:generated all 14 up to date after the main merge (os-regen-merge.sh); dispatch-gates 97/97 run with 96 exit 0. On 0b024ae9, a changeset-only delta: the changeset gates exit 0 except check-empty-changeset.
4. Accepted deviations: ① The PR edits two pending changesets it did not add, .changeset/28565-field-patch-applied.md and .changeset/28560-field-patch-slot.md. Neither has been consumed by a release on main. In each, only the code example's option changes, from tech: { label: 'Technology' } to technology: { label: 'Tech' }; no behaviour sentence changes. This is a DELIBERATE CORRECTION, so Check Changeset stays red by design. The same-head contract-review PASS is the written confirmation, and it must judge both edits. ② 28560 is a fifth copy the card did not list; it falls inside the claimed .changeset/ surface.
5. Out of scope: none. The PR body's maintainer line 2 says docs(skills): objectstack-data teaches an object extension's fieldPatches — relabel a vendor field without redeclaring it in fields #28795 already fixed the skill copy. docs(skills): objectstack-data teaches an object extension's fieldPatches — relabel a vendor field without redeclaring it in fields #28795 is still open, so that line is corrected in this record's 速读, not by a body edit. The seat sent back the new changeset's "same spelling the published skill uses" clause for the same reason, and the dev dropped it at 0b024ae9.
6. Landing: packages/spec/src/** non-test is a contract-review face, so one CONTRACT_REVIEW_TIER review is owed on this head; it also confirms ④ ①. It is handed to triage (line below). After a PASS and all checks green except the by-design Check Changeset, the seat readies and queues the PR through the relay. A head move voids this record.

Contract-review-handoff: triage — CONTRACT_REVIEW_TIER unavailable (maintainer standing word 2026-10-28T15:04Z: the Fable quota is short, so contract reviews wait for triage) · head 0b024ae

[objectstack-fleet]

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Contract-review-claim: session_01AavokzJ5DndAwitDXvKy4U · head 0b024ae90c2013c870e437ff844dd15af785c688 · 2026-11-01T22:17Z

Taken on the lane's Request: contract-review doorbell (#28646, 6564699372). Face schema (prose in packages/spec/src), at CONTRACT_REVIEW_TIER. It covers the DELIBERATE CORRECTION of the pending changesets for 28560 and 28565.

---

Generated by Claude Code

[objectstack-fleet]

triage#1 · os-zhuang · session_01AavokzJ5DndAwitDXvKy4U

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 0b024ae90c2013c870e437ff844dd15af785c688
Local-runs: none
Scope: packages/spec/src/data/object.zod.ts + examples/app-crm/src/objects/account.object.ts, packages/spec/src/data/field-customization.ts, packages/metadata-core/src/extension-field-patch-targets.ts, content/docs/references/data/object.mdx, content/docs/data-modeling/object-extensions.mdx, .changeset/28560-field-patch-slot.md, .changeset/28565-field-patch-applied.md, .changeset/28803-field-patches-example.md

Inputs read: card #28803 and its three comments; PR #28827 body, file list, the seat's ## 席位核对 (6564679875) and my claim (6564710555); the diff of merge-base 663b7c08eedd61a3a492a425ced0dbbd57676415 to head 0b024ae90c2013c870e437ff844dd15af785c688, and the compare of base 3e05ed1ec2ecce09b19a23bd51a6f14afb40caf3 to that head; the head's check-runs and the Check Changeset job log. Nothing was built, run or tested locally. Review face: schema, because packages/spec/src/data/object.zod.ts is a published non-test file.

① Derived judgments

• No code, type or export change. The object.zod.ts diff is three string edits, 4 lines out and 4 in: the error message of the options record in FieldDisplayPatchSchema (line 4838), the TSDoc @example above ObjectExtensionSchema (line 4882), and the fieldPatches .describe() text (lines 4967 to 4968). No schema key, refinement, default or export moves. The accept and refuse results of ObjectExtensionSchema are unchanged. Judged correct.
• The only generated output the describe feeds is content/docs/references/data/object.mdx. I read the describe string at head and the fieldPatches table cell at head and compared them character by character: 789 characters each, identical. The old cell matched the old describe the same way. That is a regeneration, not a hand edit. The lint job Type Check · source gates holds the check:docs step and is green on this head. No committed JSON schema carries the describe text: a grep of the head tree for Display patches of the target object hits only object.zod.ts and object.mdx; packages/spec/json-schema.manifest/*.json lists schema names only. Judged correct.
• The two-dot diff (merge base to head) and the three-dot compare (base to head) name the same 6 files with the same line counts: +18 and -8. main moved four commits past the merge base (fix(fields): a text field that declares options reads the option label in its record cell #28817, fix(platform-objects): the Setup Users list and the user record highlights stop showing the legacy role scalar #28810, fix(service-analytics)!: the draft preview refuses one column asked raw beside its own time bucket, as the live faces do #28794, fix(runtime): the customizations read attributes a contributed field and flow by one stamp rule #28807) and none of them touches any of the 6 files or examples/app-crm/src/objects/account.object.ts. The corrected sentences are judged on the same reference field main has.
• No test pins the example text or the edited error message. Every tech: { hit on the head tree is a test fixture that builds its own object (field-customization.test.ts, field-patches-derived.test.ts, and the cli, metadata-core, objectql and cloud-connection field-patch tests), plus the new changeset's quote of the old text. The phrase keyed by each option appears in no test. Nothing had to be force-edited. Judged correct.
• Clause-②: no and Metadata-protocol: none are declared on the PR body and in .changeset/28803-field-patches-example.md. Both hold: no acceptance set widens or narrows.

② Semver level

• .changeset/28803-field-patches-example.md grades @objectstack/spec at patch. Prose only, so patch is right. The changeset declares Clause-②: no and Metadata-protocol: none, matching the PR body. Its sentence quoting the old text and the new text matches the diff.
• Check Changeset is red twice on this head (runs 41356001909 and 41356296036). I read the job log of the second run. The failing step is node scripts/check-empty-changeset.mjs --base 3e05ed1ec.... It names exactly .changeset/28560-field-patch-slot.md and .changeset/28565-field-patch-applied.md under the finding: random changeset filenames collide silently across parallel agents — a round overwrote a sibling PR's minor changeset and every gate stayed green #17712 foreign changeset rule and prints the two-class remedy; this PR is the DELIBERATE CORRECTION class. The check-changeset-no-major.mjs step in the same job passes. No skip-changeset label is applied, as the rule requires. That red is the by-design one and is confirmed below.
• DELIBERATE CORRECTION, confirmed note by note. In each note exactly one line changed; every other sentence is byte-identical to the base. The two notes are still pending on main at 3e05ed1e; no Version Packages release has consumed them.
  • .changeset/28560-field-patch-slot.md, line 19, inside the objectExtensions code example. Before: options: { tech: { label: 'Technology' } },. After: options: { technology: { label: 'Tech' } },. Is the corrected line true on this tree? Yes. crm_account.industry in examples/app-crm/src/objects/account.object.ts is an inline Field.select whose options are technology (default, label Technology), finance, healthcare, retail, other. The key technology names a stored value; the key tech named none. The note's own sentence two bullets down, that options is a map from an option's stored value to that option's patch, is now shown by the example instead of contradicted by it. No behaviour sentence changed.
  • .changeset/28565-field-patch-applied.md, line 20, inside the objectExtensions code example. Before: industry: { label: 'Sector', inlineHelpText: 'The market the account sells into', options: { tech: { label: 'Technology' } } },. After: industry: { label: 'Sector', inlineHelpText: 'The market the account sells into', options: { technology: { label: 'Tech' } } },. Is the corrected line true on this tree? Yes, by the same field. This note is the one that says a patch naming an option value the field does not have is refused at install and by os validate, so before the correction the note's own example was refused by the rule the note announces. No behaviour sentence changed.
• Both edits sit inside the claimed root .changeset/ surface. The 28560 edit is a fifth copy the card did not list; leaving it would republish a refused example, and the card's Wanted says to change a pending copy while it is still unreleased.

③ Boundary flags

• Dev open_questions, who confirms the two pending changeset corrections: answered. Option A. This same-head PASS record is the written confirmation, per landing-operations line 16 (a correction beyond release text alone is confirmed by the at-tier same-head PASS, note by note). The seat's ACCEPT item 4 already expected this.
• Dev deviation, editing .changeset/28560-field-patch-slot.md: accepted, see ②.
• PR author user.login is objectstack-fleet[bot], so no author note is owed.
• PR body, maintainer summary line 2, says the skill copy was already fixed in docs(skills): objectstack-data teaches an object extension's fieldPatches — relabel a vendor field without redeclaring it in fields #28795. docs(skills): objectstack-data teaches an object extension's fieldPatches — relabel a vendor field without redeclaring it in fields #28795 is still open at this reading. The seat corrected that line in its record and the dev dropped the matching clause from the new changeset at this head. On the head tree skills/** carries no fieldPatches example at all, so no skill copy disagrees with this PR; docs(skills): objectstack-data teaches an object extension's fieldPatches — relabel a vendor field without redeclaring it in fields #28795 adds one spelled technology: { label: 'Tech' }, which agrees. Not a must-fix.
• The seat's record says 7 files; the PR has 6. Cosmetic, no bearing on the verdict.
• Open card spec(object-extension fieldPatches): refuse a field both replaced in fields and patched in fieldPatches of one extension, key option patches by the option-value grammar, carry the whitelisted aliases, and derive the describe's member list #28655 (refuse a field both replaced in fields and patched in one extension) is not live on this tree; the examples do not mix the two anyway, see ④.
• Check-runs on the head at 2026-11-01T22:23Z: red, 2, both Check Changeset, the by-design correction refusal confirmed above; pending, 29: Build Core, Build & E2E UI, Live E2E UI, Type Check UI, Type Check · consumer gates, Type Check · workspace, Test Core 1 to 8, Test UI 1 to 16; everything else green or an expected skip (Governed Surface Queue Guard, Spec property liveness, Lint & Repo Gates, Type Check · source gates, Build Docs, PR-body guards all green). Landing still needs the pending set green; that is the lane's gate check, not a finding against the diff.
• The docs guide copy writes inlineHelpText: 'The market the account sells into.' with a trailing period; the other copies have none. Pre-existing and not part of this card. Noted only.

④ Design judgment

packages/spec/src/data/object.zod.ts, judged against the field it cites and the rules it teaches.

• The stored value exists. examples/app-crm/src/objects/account.object.ts line 27: { label: 'Technology', value: 'technology', default: true }. Same on main at 3e05ed1e.
• The patched label is a real change. The field's label for technology is Technology; the example patches it to Tech. The old example patched tech to Technology, which would have been a no-op label even if the key had existed.
• The example passes the fieldPatches rules on this tree. Field keys used: label and inlineHelpText; both are in BASE_FIELD_DISPLAY_WHITELIST.field (packages/spec/src/data/field-customization.ts line 77: label, description, inlineHelpText, placeholder). Option key used: label; it is the whole of BASE_FIELD_DISPLAY_WHITELIST.option (line 79). No option value is set, so the value refusal in FieldOptionDisplayPatchSchema does not fire. options is a map, not an array, so the array refusal whose message this PR edits does not fire. No __c key, so refuseExtensionFieldPatchKeys does not fire. At the server, fieldPatchTargetMisses (packages/metadata-core/src/extension-field-patch-targets.ts lines 92 to 132) finds the field, sees no picklist, and finds technology in the field's option values, so no option miss; tech produced one.
• The patch is not mixed with fields. The TSDoc example declares fields: { sales_stage: ... } and patches industry, two different fields. The describe example, the error message, the two docs copies and the two pending notes carry no fields at all.
• Every copy agrees. Eight sites on the head tree spell the option technology: { label: 'Tech' }: object.zod.ts lines 4838, 4882 and 4967 to 4968; content/docs/references/data/object.mdx line 579; content/docs/data-modeling/object-extensions.mdx line 113; .changeset/28560-field-patch-slot.md line 19; .changeset/28565-field-patch-applied.md line 20; and .changeset/28803-field-patches-example.md line 10 quotes the old and the new text correctly. No tech: { remains outside test fixtures.
• Verdict on the face: the describe, the TSDoc and the refusal text now teach an example that os validate and the install door accept, and they still say what the schema does. No must-fix.

Implemented-by: claude/project-thread-7jsd3z
Reviewed-by: session_01AavokzJ5DndAwitDXvKy4U

VERDICT: PASS

---

Generated by Claude Code