[objectstack-fleet]

Requested by Warren · project thread
Fixes #29601

Clause-②: no
Metadata-protocol: none

维护者速读

• 改了什么:ADR-0146 D8 第 2 条里「the export of X's app surface」改为点名配对包的导出 GET /packages/X.custom/export(指向第 3 条),并写明它永远不是基础包的导出 GET /packages/X/export;第 2 条首句与「Decided here」[WIP] Fix error in step four of the action run #5 随之收窄;Refs 加一条带日期的出处。只改措辞,+12/−6。
• 为什么改:原句可读成基础包导出,与第 3 条「基础包导出原样、配对包导出带 overrides 声明」矛盾;packages: the base export carries a published dashboard or report override from its .custom pair, where ADR-0146 D8 item 3 says the base export is unchanged #29563 的开发把它列为缺陷,PR fix(runtime): the base's package export leaves out its pair's dashboard and report overrides (#29563) #29597 和 getMetaItemsPerPackage 的注释都按配对包导出理解。
• 风险与回滚:无代码、无 schema;不改任何裁决,Status 行不加修订条目(出处记在 Refs);回滚即还原这两个 commit。
• 先看:docs/adr/0146-studio-designs-setup-administers-one-editor-per-item-paired-customize-package.md D8 第 2 条,其次「Decided here」[WIP] Fix error in step four of the action run #5 与 Refs。
• 两分钟验证:git diff origin/main -- docs/adr/0146-*.md,预期只有 D8 第 2 条、「Decided here」[WIP] Fix error in step four of the action run #5 与 Refs;pnpm check:adr-links && pnpm check:adr-anchors 预期绿。
• 需要你做的:Tier H(docs/adr/**),席位合同审查 PASS(6595888270,delta 6596113742)且已 ACCEPT;需要 GOVERNED_APPROVERS 的 APPROVED 后落地。

What changes

One sentence of ADR-0146 D8 item 2. The list of reads that serve a declared override named "the export of X's app surface". It now names the export it means: the pair's export, GET /packages/X.custom/export, with a pointer to item 3, and says it is never GET /packages/X/export, the base's own export, which item 3 keeps unchanged. The contract review's should-fixes (6595888270) moved two more spots in the same file: item 2's lead now says the override is served "in every read but the base's own export (item 3)", and Decided here #5 is scoped to the four reads under the base's address, with the override travelling in the pair's export, never the base's. Refs gains a dated entry naming #29601 and this PR. No code, no schema, no other ADR.

The reading taken is the seat's (dispatch order, claim 6594501078): item 3 already splits the two routes, PR #29597 implemented that split and read item 2 this way, and the getMetaItemsPerPackage docblock in packages/metadata-protocol/src/protocol.ts reads it the same way. The other reading the card lists, a whole-surface export serving effective documents, is #28914's tenant ontology export, which is unlanded and unnamed in ADR-0146; naming it there would be a new decision, so this edit does not.

No amendment entry on the Status line: the record's own convention reserves **Amended in place** for rulings that change a decision (ruling records on decision cards). This clarification reverses nothing and picks the reading item 3 already fixes; the card and this PR are its provenance.

Why

Read literally, item 2 said the base's export serves the override, while item 3 says the base's export carries the base unchanged. Code on main at c76c0e04 agrees with item 3 and with the clarified item 2: ownDocumentsOf in packages/runtime/src/domains/packages.ts drops or replaces an item that declares it overrides the exported package, so GET /packages/BASE/export serves the base's own documents and the pair's export carries the override with its overrides block. No code on main contradicts the reading. The whole-deployment export GET /packages/export that #29755 landed on main (b3d14150) while this PR was open answers each package exactly as its per-package export, the pair after its base, so it agrees with item 3 and is not named here (delta review 6596113742: a docs-only follow-up).

Tests

Docs-only; no test changes. The gate families derived for the diff ran locally (Verification).

Acceptance notes

• Naming GET /packages/export (packages: a tenant exports its whole ontology — every installed package, its .custom pair and environment customizations — as installable packages for the open runtime (os package export + the read it calls) #28914, landed by feat(runtime,client,cli): export a deployment's whole ontology as installable packages in one read (GET /packages/export, os package export) #29755) in ADR-0146 is a new decision and goes to a docs-only follow-up card, not this clause.
• skip-changeset: docs/adr/** publishes nothing (the fast lane of the changeset rule).

Verification

Head 71e8cc17 (two commits on origin/main c76c0e04; the second answers the contract review's should-fixes). The same 21 derived gates re-ran at 71e8cc17: 21 run, all exit 0, 0 NOT-MEASURED (os-dev-report 6596068319).

dispatch-gates --commands (21) and --ran reconciliation, first run at 954c3bd

node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack derived 21 commands from the diff (1 path, +3/-1). Every one ran from the worktree with its exit code captured before any pipe; all 21 exit 0. Prerequisite for check:doc-formula-expressions: @objectstack/formula and @objectstack/lint built under the verify lock (VERDICT command-exit 0, held 170 s).

node scripts/pm/dispatch-gates.mjs --ran ran.list --repo objectstack-ai/objectstack: 21 derived, 21 run, 0 NOT-MEASURED, 0 UNRUN (exit 0).

Exit 0: check-adr-links (+ --self-test), check-closing-keyword-parity (+ --self-test), check-comment-mask-corpus, ci/plan-test-core-shards --self-test, lint check:doc-formula-expressions, check:adr-anchors, check:adr-index, check:adr-symbol-anchors, check:ci-filter-parity, check:closed-ceremony-text, check:cross-package-test-inputs, check:doc-authoring, check:driver-memory-census, check:gitlink-declared, check:nul-bytes, check:pm-governed-merges, check:pm-prior-rulings, check:refd-timer-probe, check:watch-hint-literal.

Repo-level scans (pnpm lint, the full check:* farm) are CI's runs on this PR; read its check-runs.

Governed surface (docs/adr/**, Tier H): this PR stays a draft; it lands only after an authorized APPROVED review from GOVERNED_APPROVERS, following the owning seat's contract-tier review and ACCEPT. No reviewers requested by the dev.

Human Minimum

• 1. Schemas start as Zod; packages/spec holds no runtime logic. (No packages/spec change.)
• 2. Every name I added uses the casing for its kind. (No names added.)
• 3. No formatter ran; pnpm lint is CI's repo-level run (no code touched).
• 4. One issue, one branch off an up-to-date main, one pull request.
• 5. A change that publishes carries a changeset: docs-only, publishes nothing; both declaration lines are filled in above.
• 6. packages/spec touched: n/a.
• 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 the os-dev house format).
• The Before Submitting list in CONTRIBUTING.md is done.

🤖 Generated with Claude Code

https://claude.ai/code/session_01JQwxbYeS2U1HKXS13SzfKe

---

Generated by Claude Code

[objectstack-fleet]

domain:server#3 · os-warren · session_01JQwxbYyp6wSRcywGFK2L4A

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 954c3bd708f63bad27e7eb1e1dd8a8482cf28174
Local-runs: none
Scope: docs/adr/0146-studio-designs-setup-administers-one-editor-per-item-paired-customize-package.md + packages/runtime/src/domains/packages.ts (declaresOverrideOf, ownDocumentsOf, assemblePackageManifest, the GET /packages/:id/export route), packages/metadata-protocol/src/protocol.ts (the getMetaItemsPerPackage docblock), packages/runtime/src/domains/packages-export-base-unchanged-per-type.integration.test.ts (the #29597 pin)

Reviewer: an at-tier subagent adopted by this seat. Inputs: card #29601 and every comment, #29563 and its report 6588016343, merged PR #29597, #28914, ADR-0146, PR #29791's body, file list and net diff (one hunk, +3/−1), the head's check-runs (18 success, 5 unreachable-from-ADR lanes still running at review time, none failed).

① Derived judgments

• The pair's export GET /packages/X.custom/export serves the overriding item with its overrides block: holds (assemblePackageManifest(deps, 'X.custom', …) → ownDocumentsOf keeps it since declaresOverrideOf(item, 'X.custom') is false; overrides survives cleaning; pinned for view/dashboard/report by the fix(runtime): the base's package export leaves out its pair's dashboard and report overrides (#29563) #29597 test).
• The base's export GET /packages/X/export serves the base unchanged: holds (ownDocumentsOf drops overrides.package === 'X' items and restores the registry's own artifact; pinned).
• Item 2 now agrees with item 3 and with the code's own comments (assemblePackageManifest, the getMetaItemsPerPackage docblock).
• No ontology export is named anywhere under packages/runtime/src, packages/rest/src or packages/metadata-protocol/src; the retired phrase occurs nowhere else in the ADR.

② Semver level

none: docs/adr/** only, skip-changeset right; Clause-②: no, Metadata-protocol: none right and line-initial. Check Changeset green.

③ Boundary flags

• Tier H posture correct (draft, no auto-merge, no reviewer yet).
• No Status-line amendment: right by the rule (this reverses nothing; ADR-0146 reserves Amended in place for rulings). But the in-place rewrite carries no dated card marker, while every recent ADR edit on main does (e.g. 67103cc, 325944b) and ADR-0146's own Refs lists its earlier wording-fix PR (docs(adr): ADR-0146 — anchor the four load-bearing symbols, count two Setup pages, drop the epic label snapshot (S13–S15) #28145). Should-fix 1: add one dated provenance marker (a Refs line naming docs(adr-0146): D8 item 2's "the export of X's app surface" can be read as GET /packages/X/export, which item 3 says carries the base unchanged; name the export it means #29601 / PR docs(adr-0146): D8 item 2 names the pair's export, GET /packages/X.custom/export, never the base's #29791 beside docs(adr): ADR-0146 — anchor the four load-bearing symbols, count two Setup pages, drop the epic label snapshot (S13–S15) #28145, or a parenthetical at the sentence).

④ Design judgment

• No new decision; packages: a tenant exports its whole ontology — every installed package, its .custom pair and environment customizations — as installable packages for the open runtime (os package export + the read it calls) #28914 is not foreclosed. Its Wanted §1 composes per-package exports including the .custom pair, so it depends on exactly the split this sentence states; the appositive "the pair's export" keeps the sentence from describing a future bundle.
• "never GET /packages/X/export" is consistent with item 1 (addressing only) and with the heading's agreement claim, which is agreement on the document served; the pair's export carries that document.
• Should-fix 2: item 2's bold lead "Within X's address, a declared override is served over the base's document" and "The read that serves X's name" are now over-general against an enumeration that includes a read under X.custom's address (and the base's export, inside X's address, does not serve the override). Qualify the lead or the enumeration, e.g. "the reads that serve X's name or carry X's customizations".
• Should-fix 3: "Decided here where the rulings were silent" [WIP] Fix error in step four of the action run #5 still says a declared override is served within the base's address "for all five read kinds", which overstates for export (item 3). Point [WIP] Fix error in step four of the action run #5 at item 3 or scope it to the four reads under X's address.

No must-fix.

Implemented-by: claude/project-thread-x37kcs
Reviewed-by: session_01JQwxbYyp6wSRcywGFK2L4A

VERDICT: PASS

[objectstack-fleet]

domain:server#3 · os-warren · session_01JQwxbYyp6wSRcywGFK2L4A

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 71e8cc1774bc37bb5b942b3cb3d4469b31d5981d
Local-runs: probe — node scripts/check-adr-links.mjs ran once by accident, on the clone's origin/main tree and not on the head, so nothing below relies on its result
Scope: docs/adr/0146-studio-designs-setup-administers-one-editor-per-item-paired-customize-package.md + packages/runtime/src/domains/packages.ts (assemblePackageManifest, ownDocumentsOf, assemblePackageExportBundle / GET /packages/export as #29755 landed them at b3d14150), .changeset/28914-package-export-whole-deployment.md, packages/spec/src/api/package-lifecycle.zod.ts (PackageExportBundleSchema description)

VERDICT: PASS

This is a delta review of 954c3bd7..71e8cc17: one commit, no force push, +10/−6 in the one ADR file. The reviewer is the one who wrote record 6595888270, and this review judges only its three should-fixes and whether any new must-fix appeared.

① Derived judgments

• Should-fix (1), Refs provenance: landed correctly. Refs now carries a dated entry for card docs(adr-0146): D8 item 2's "the export of X's app surface" can be read as GET /packages/X/export, which item 3 says carries the base unchanged; name the export it means #29601 and PR docs(adr-0146): D8 item 2 names the pair's export, GET /packages/X.custom/export, never the base's #29791. It sits beside the docs(adr): ADR-0146 — anchor the four load-bearing symbols, count two Setup pages, drop the epic label snapshot (S13–S15) #28145 wording-fix entry and says what moved, following the record's own convention.
• Should-fix (2), the D8 item 2 lead: landed correctly. The lead now says "in every read but the base's own export (item 3)". The list now reads "The reads that serve X's name or carry X's customizations", and the plural verb agrees. The pair's export now falls under the "carry X's customizations" limb, which keeps the lead consistent with item 3 and with ownDocumentsOf on main. The reviewer tried to break the lead and found no must-fix.
• Should-fix (3), Decided-here [WIP] Fix error in step four of the action run #5: landed correctly. [WIP] Fix error in step four of the action run #5 is now scoped to the four reads under the base's address (package-less, named, layered and list), and says the override travels in the pair's export, never the base's (D8 items 2 and 3). The remaining "five reads" phrasings (the pin at L440, N-10 at L632) say only that the reads agree on the document served, which the pair's export shares, so they stay correct.

② Semver level

Unchanged: none, skip-changeset, Clause-②: no, Metadata-protocol: none; the PR still touches only docs/adr/**.

③ Boundary flags

• The Tier H posture is unchanged: the PR is still a draft with no reviewer requested, and Governed Surface Queue Guard is green.
• At review time, CI showed no failures; 9 runs were still in progress, including Lint · docs & code guards, the run that reads the ADR hunk. The seat waits for green before ACCEPT.

④ Design judgment

Naming the whole-deployment export GET /packages/export (#28914, landed by PR #29755 at b3d14150) is a follow-up, not a must-fix here.

• assemblePackageExportBundle returns, for each package, exactly what GET /packages/:id/export returns, in boot order, so a pair comes after its base. The spec description and the changeset say the same.
• The bundle therefore adds no new read of X's name with a different answer, and D8 items 2 and 3 pass through it unchanged. The edited sentences stay true on b3d14150.
• Naming the bundle in ADR-0146 is the new decision this wording-only card excluded.
• The follow-up should be a docs-only card with a one-line pointer in D8 item 3 or in Refs: "the whole-deployment export lists each installed package's own export unchanged, the pair after its base".

There is no must-fix. All three should-fixes from the prior record landed as asked, so this review stops here.

Implemented-by: claude/project-thread-x37kcs
Reviewed-by: session_01JQwxbYyp6wSRcywGFK2L4A

VERDICT: PASS

[objectstack-fleet]

domain:server#3 · os-warren · session_01JQwxbYyp6wSRcywGFK2L4A

维护者速读

• 改了什么：只改 docs/adr/0146-studio-designs-setup-administers-one-editor-per-item-paired-customize-package.md（+12/−6）。
  • D8 第 2 条原写「the export of X's app surface」，现点名配对包的导出 GET /packages/X.custom/export，并写明它永远不是基础包的导出 GET /packages/X/export（见第 3 条）。
  • 第 2 条首句改为「除基础包自己的导出外的每个读取」。
  • 「Decided here」[WIP] Fix error in step four of the action run #5 收窄到基础包地址下的四种读取。
  • Refs 加一条带日期的出处（docs(adr-0146): D8 item 2's "the export of X's app surface" can be read as GET /packages/X/export, which item 3 says carries the base unchanged; name the export it means #29601 / docs(adr-0146): D8 item 2 names the pair's export, GET /packages/X.custom/export, never the base's #29791）。
• 为什么改：原句可以读成「基础包导出带覆盖」，与第 3 条「基础包导出原样、配对包导出带 overrides」矛盾。PR fix(runtime): the base's package export leaves out its pair's dashboard and report overrides (#29563) #29597 和 ownDocumentsOf 都按配对包导出实现。
• 风险与回滚：纯文字改动，不发布、不改行为，也不改任何裁决，Status 行不动。回滚即 revert。
• 先看：D8 第 2 条，其次「Decided here」[WIP] Fix error in step four of the action run #5。
• 两分钟验证：pnpm check:adr-links && pnpm check:adr-anchors，预期绿。
• 需要你做的：本 PR 属 Tier H（docs/adr/**），请 GOVERNED_APPROVERS 之一批准，批准后由席位落地。
• 席位意见：建议批准。
  • fable 达档复核两轮：PASS 6595888270，delta PASS 6596113742。三条 should-fix 都已落在本 head。
  • 按你的常设档位令以 fable + xhigh 派发。
  • 后续：feat(runtime,client,cli): export a deployment's whole ontology as installable packages in one read (GET /packages/export, os package export) #29755 新落地的整部署导出 GET /packages/export 逐包照搬单包导出，与本条一致。ADR-0146 还没点名它，点名属于新决定，另开纯文档卡处理，不在本 PR。

席位核对

ACCEPT · domain:server#3 · head 71e8cc1774bc37bb5b942b3cb3d4469b31d5981d · 2026-11-03T05:28Z · governed docs/adr/** (Tier H: check-governed-merges --pr 29791 reports 1 of 1 path governed, 18 lines). CONTRACT_REVIEW_TIER review PASS on this head (6596113742, delta of 6595888270). This is the final four-piece hand-off to the approvers.

1. Shape: draft, base main, Fixes #29601, Clause-②: no, Metadata-protocol: none, skip-changeset (docs-only fast lane). 1 file, +12/−6. The seat refreshed the PR body to this head (relay issue_patch, read-back identical); this also dropped the stale "nothing names an ontology export" sentence.
2. Premise measured: on origin/main c76c0e04, item 2 could be read as the base's export, against item 3. ownDocumentsOf (packages/runtime/src/domains/packages.ts) serves the base unchanged and puts the override in the pair's export. On b3d14150, assemblePackageExportBundle (feat(runtime,client,cli): export a deployment's whole ontology as installable packages in one read (GET /packages/export, os package export) #29755) composes the per-package export verbatim, so the reading still holds.
3. Gates on this head: check-runs pinned to 71e8cc17: 45 total, 26 success, 19 skipped, 0 failed, including Lint · docs & code guards (check:adr-links, check:adr-anchors, check:doc-authoring). mergeable_state is clean. Dev: 21/21 derived gates exit 0 on 954c3bd7 and again on 71e8cc17 (os-dev-report 6596068319).
4. Prose checked against the diff: the ADR text is the reviewed face (records above). The PR body's What changes, Why and Acceptance notes match the diff on this head.
5. Out of scope: naming GET /packages/export in ADR-0146 is a new decision. It gets a docs-only follow-up card from the seat, not this PR. Skills-docs: not owed.
6. Landing: Tier H ⇒ relay pr_request_reviewers to GOVERNED_APPROVERS (os-zhuang, hotlong) and needs-user-decision on the PR in the same act; the PR stays draft. On an approver's APPROVED: clear the label, pr_ready + automerge_enable, follow to MERGED, landing record on docs(adr-0146): D8 item 2's "the export of X's app surface" can be read as GET /packages/X/export, which item 3 says carries the base unchanged; name the export it means #29601.