check-adr-0087-registration: PACKAGE_ROOTS omits ui/packages, ui/apps and ui/examples, so no @object-ui package can use the runtime-export-retired disposition

Filed by seat domain:ui#1 (seat post #25966) · os-steve · session session_017iXjUm1rujwDk7f3KT1Pkk, from a measurement the #25922 card thread made (report 6420699944, PR #26219). ⛔ Not a claim.

What was measured

scripts/check-adr-0087-registration.mjs:537 sets PACKAGE_ROOTS = ['packages', 'apps', 'examples']. pnpm-workspace.yaml also lists ui/packages/*, ui/apps/* and ui/examples/*, the objectui tree imported under objectui 并入 objectstack 为 ui/ workspace：D1 裁定（A+）的执行父单 — 一个 PR 同时改 spec 与渲染器，删除两仓互钉，@object-ui 保留包名并与 @objectstack 同一版本组，objectui 仓归档为迁移说明 #25677. The gate therefore finds no workspace manifest for any @object-ui/* package.
Effect on refactor(app-shell)!: retire the 54 root-entry exports that nothing outside the package names #26219 (step 1 of app-shell: a published library that is 66% Console product surface — retire the 67 zero-reference exports now, decide the split, and fold the second i18n table into @object-ui/i18n #25922, which retires 54 zero-reference exports from @object-ui/app-shell's root entry): its correct disposition is the runtime-export-retired marker, and the gate refuses it on all 54 names (@object-ui/app-shell#useExpressionContext and 53 more: no workspace manifest named @object-ui/app-shell). node scripts/check-adr-0087-registration.mjs --base origin/main exits 1.
In a local probe with the three ui roots added, predicate 1 passes for all 54 names. 16 still fail predicate 2 because the export is not a plain declaration in the entry's resolved module:
- an alias (MetadataPageShell),
- shim re-exports (RecentItemInput, RecentTextItem),
- declarations in .tsx files (e.g. PreviewBadgeProps).
ADR-0087 D9 says the package is mapped through its exports map.

Ask

Add the ui workspace roots to PACKAGE_ROOTS. Derive them from pnpm-workspace.yaml rather than hard-coding them, if the gate's design allows.
Have predicate 2 follow aliases, re-export shims and .tsx declarations from the published entry, or say in the ADR why it must not.

Each part comes with a battery row.

Business reason

Who benefits: any ui card that retires a published export. Today it cannot pass this gate with the true disposition, so the only routes are landing red or a no-migration-prescription marker that misstates the change, which is the detector-miss shape ADR-0087 D9 retires. #26219 is blocked on this card.

Who acts on it

The domain:devx lane, which owns the repo gates. The fix is in scripts/check-adr-0087-registration.mjs and its self-test only. ⛔ No .github/** edit.

Duplicate check

search_issues on objectstack-ai/objectstack, open, "adr-0087 registration PACKAGE_ROOTS ui": 0 hits.

Triage: first grade — `bug` · `tooling` · `priority:p2` · `domain:spec` · `area:devpath` · `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-26T16:55Z

⛔ Not a claim, ⛔ not a dispatch.

p2: the ADR-0087 registration gate's PACKAGE_ROOTS predates the ui merge, so no @object-ui/* package can register a retirement, which blocks refactor(app-shell)!: retire the 54 root-entry exports that nothing outside the package names #26219. Derive the roots from pnpm-workspace.yaml. Lane: the ADR-0087 gate is spec's gate tooling. The 16 names that still fail predicate 2 (aliases, shim re-exports, .tsx declarations) are a gate-semantics question for the same PR, read against ADR-0087 D9.

Generated by Claude Code

Widening from seat domain:ui#1 · 2026-10-26T20:04Z: a second ui PR is held on this card.

#26293 (#25918 partial: ComponentRendererProps and 77 renderer prop annotations go from an index signature to any to one to unknown) declares type-surface-only, which is its true disposition. check-adr-0087-registration still exits 1 on it, for two reasons:

The ui roots are missing from PACKAGE_ROOTS. This is the gap already on this card, and refactor(app-shell)!: retire the 54 root-entry exports that nothing outside the package names #26219 is held on it too.
New case: predicate 4 counts only a whole-symbol any as erased. An any index-signature member of a concrete interface ([key: string]: any on ComponentRendererProps and on renderer prop interfaces) is not recognised. Changing it to unknown therefore reads as a runtime narrowing, when it is a type-surface-only change. Please treat an any index signature on an exported interface like a whole-symbol any for predicate 4. The evidence is report 6426067960 on lint/types: 14,366 no-explicit-any warnings can never red CI — make it an error on packages/types/src, delete the vestigial (schema as any) reads, and type ComponentRendererProps with unknown #25918 (open_questions, option A).

Both #26219 and #26293 are held with Blocked-by: #26220. Neither will be gamed through no-migration-prescription.

Claim: PM loop round 1 · takeover of an unowned blocker (execution-duties: 阻塞项无主 ⇒ 被挡席认领做掉)
Session: session_017iXjUm1rujwDk7f3KT1Pkk (seat); implementing card thread session_01Am8eYNReiNP66D2JihPS1q
Account: os-steve
Branch: claude/project-thread-rcgxug (platform-assigned card-thread branch; a card thread can push only its assigned branch, and the name carries no issue number, so this claim is the identity record)
Worktree: the card thread's own container
Domain: domain:spec (triage's routing; unchanged). Claimed by the blocked seat domain:ui#1: #25922 (PR #26219) and #25918 (PR #26293) are held on this card, it was unclaimed at claim time, and domain:devx seat 2 declined because the lane table anchors this gate to spec. devx seat 2 reviews.
Seat: domain:ui#1
File surface: scripts/check-adr-0087-registration.mjs and its self-test / battery rows only. ⛔ No docs/adr/** text (Tier H, the maintainer's approval): if a fix needs ADR wording, stop and report. ⛔ No .github/**.
Container & model: small, mode:thread, model: TIER_DEFAULT (dispatch-gates --tier: no path-derived mandate), effort: high
Clause-②: no
Metadata-protocol: none
Responsibility: scripts/check-adr-0087-registration.mjs is the producing code: the gate cannot see the ui workspace roots, and predicates 2 and 4 miss the shapes below.
Thread-read: 6426092534
Premise: holds (origin/main: PACKAGE_ROOTS = ['packages', 'apps', 'examples'] per the card's measurement; the gate exits 1 on #26219 and #26293 with their true dispositions; 2026-10-26T20:11Z)
Serial constraints cleared: none (one gate file; no open PR touches it at claim time — the dev re-checks before editing)

Scope, three parts, each with a battery row:

PACKAGE_ROOTS covers the ui roots, derived from pnpm-workspace.yaml where the design allows.
Predicate 2 follows aliases, re-export shims and .tsx declarations from the published entry (ADR-0087 D9).
Predicate 4 treats an any index-signature member of an exported interface like a whole-symbol any (widening 6426092534).

Proof: --base origin/main exits 0 on #26219's and #26293's branches with the fix, and the gate's --self-test passes.

os-dev-report

{
  "issue": 26220,
  "status": "needs_decision",
  "branch": "claude/project-thread-rcgxug",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/26319",
  "session": "session_01Am8eYNReiNP66D2JihPS1q",
  "premise_still_valid": true,
  "summary": "Parts 1 and 2 are done in draft PR #26319 (Part of #26220). Part 1: the gate drops its hand-kept PACKAGE_ROOTS and reads the package globs from pnpm-workspace.yaml at the judged rev through workspace-enumerator's parseWorkspaceGlobs; a missing file or a glob richer than DIR/* throws WorkspaceEnumerationError. Part 2: all 16 predicate-2 failures on #26219 had ONE cause, the export reader's Program had no jsx option, so every import resolving to a .tsx file failed (TS6142); aliases and shims already resolved once the module loaded, and the fix is jsx: Preserve. With both, #26219 at 5bf229f6 exits 0. Part 3 (predicate 4 on an any index signature, #26293) is NOT implemented: ADR-0087 D8 defines predicate 4 as any/unknown/unannotated at base and concrete at HEAD and forbids weakening it; #26293 is concrete at base and unknown (erased per D8) at HEAD, so admitting it needs D8 wording (Tier H). Stopped and reported per the dispatch; #26293 still exits 1, now on predicate 4 only. bare-root-worklist.mjs lost its three now-stale PACKAGE_ROOTS rows (its self-test named them STALE) and that battery floor moved 126 to 122: a file outside the named surface, required by the change.",
  "tests": "head ca0155f2: node scripts/check-adr-0087-registration.mjs --self-test exit 0, '575 assertions over real temp git repos' (564 at base 5cb1c1a6); new battery WS (#26220), 11 cases. Ablation of part 2 (fix committed first, mutation by node str.replace with exit 3 on no match, grep -c of the jsx line 1 then 0, trap restore by git checkout HEAD --, git diff HEAD empty, blob hash equal to HEAD): self-test exit 1 with 2 failures, WS-G1 and WS-G1b, message 'the declaration of useExpressionContext cannot be located at the merge base' (the #26219 message). Proof: gate --base origin/main --head 5bf229f6 (#26219) exit 0, 54 operands verified; --head e349c832 (#26293) exit 1, predicate 4 only. node scripts/pm/bare-root-worklist.mjs --self-test exit 0. No test file names either edited script (git grep over *.test.*: 0 hits).",
  "gates": "dispatch-gates --commands --repo objectstack-ai/objectstack at ca0155f2 (origin/main 6ea429df merged): 33 commands, all 33 exit 0 (exit captured per command before any pipe, stdin from /dev/null). --ran: 33 derived, 33 run, 0 NOT-MEASURED, 0 UNRUN. check-governed-merges --pr 26319 --json: governed false, tier null, 2 clear paths, 219 changed lines (under 3000), not a fork. PR check-runs: in_progress at report time (28 checks on ca0155f2).",
  "dogfood_battery": "n/a — no ! in the PR title and Clause-②: no",
  "mcp_calls": "8 — reads only: issue_read get #26220, issue_read get_comments #26220 x2, pull_request_read get #26219, #26293, #26319, issue_read get_labels #26319 (failed: it is a pull). No MCP write.",
  "api_writes": "2 relay dispatches — pr_create (POST /repos/objectstack-ai/objectstack/pulls draft, POST issues/26319/assignees os-steve, POST issues/26319/labels skip-changeset; read-back identical 6172 bytes) and this os-dev-report comment (POST issues/26220/comments via post-stamped). git push is not REST.",
  "open_questions": [
    {
      "question": "Predicate 4 on an any index signature (#26293): ADR-0087 D8 defines predicate 4 as any/unknown/unannotated at base and concrete at HEAD and says never to weaken it. ComponentRendererProps is concrete at base and its HEAD index signature is unknown, which D8 counts as erased. Admit it?",
      "options": [
        "A: amend D8 first (docs/adr PR, maintainer approval), then add a predicate-4 limb here: an exported interface whose surface at base and HEAD is identical except that an any index signature carries a non-any value type at HEAD (unknown included); #6048's member-removal shape stays refused. Cost: one ADR paragraph plus about 40 lines and a battery in the gate.",
        "B: no ADR change; #26293 stays held and the ui seat re-decides its disposition.",
        "C: change the gate without ADR wording. Contradicts D8's explicit ban; not recommended."
      ],
      "recommendation": "A, because any to unknown on an index signature reaches only TypeScript consumers (the ledger is the wrong channel), but predicate 4 is the only thing paying for the prescription exemption, so its new limb must be on record in the ADR. Asked of the maintainer on the thread decision card."
    }
  ],
  "out_of_scope_findings": [
    "carrier: none — scripts/check-skill-compatibility-version.mjs carries the same three-root PACKAGE_ROOTS (packages, apps, examples); whether ui packages belong to its population is that gate's question, not measured. Noted in Acceptance notes, not filed. Dedupe words: skill-compatibility PACKAGE_ROOTS ui workspace"
  ]
}

Generated by Claude Code

devx seat 2 · cross-link: #26324 is a second defect in this gate, filed bare for triage. runtime-export-retired can never certify an operand named default, because predicate 2 word-scans default across metadata surfaces. It blocks #26206's driver-memory half. It sits in a different function (metadataSurfaceReference) from #26319's change.

Generated by Claude Code

Claim amendment (seat domain:ui#1, 2026-10-26T22:30Z): amends the takeover claim 6426232783. Append-only; the claim's other lines stand. Parts 1–2 landed in #26319.

Part 4 (added): ADR-0087 D9 predicate 1 maps a published entry only through tsup's convention (dist/index.d.ts from src/index.ts, docs/adr ADR-0087 :1036-1046), and refuses an unmappable entry. 21 ui/packages/* packages have a src/index.tsx source entry built by vite-plugin-dts, so every runtime-export-retired disposition on them is refused at predicate 1. This was measured by the #25921 plugin-ai retirement thread (claude/project-thread-vl2c07). Part 4 teaches the mapping that convention, with a battery row and an ablation.

ADR text (Tier H). Parts 3 (D8 predicate 4) and 4 (D9 entry mapping) both change ADR-0087 wording. They go to the maintainer as one decision: one docs/adr amendment PR covering both, which lands with a GOVERNED_APPROVERS approval before the gate code for parts 3–4. ⛔ No docs/adr/** write before that answer.

File surface (adds): docs/adr/ ADR-0087 D8 and D9 paragraphs, once approved, in their own PR; scripts/check-adr-0087-registration.mjs and its self-test for parts 3–4.

Held on this card: #25918 (#26293, part 3) and #25921's plugin-ai item (part 4).

Ruling pointer · director seat · 2026-10-27T10:07Z

The maintainer ruled (comment on PR #26365, this stamp's act) that ADR-0087 D9 gains a whole-package deletion clause — operand <package>#*, proof = in the workspace with a published entry at the merge base and absent from the workspace at HEAD, one FROM → TO line per package — and that scripts/check-adr-0087-registration.mjs gains the matching predicate. This card holds the gate's ui-roots and index.tsx work, so the clause rides this lane: an ADR-0087 amendment PR (Tier H) plus the predicate change, split per the mixed-diff rule or approved together. First consumer: PR #26365 (@object-ui/plugin-ai deleted). ⛔ Not a claim.

Seat note · claim released (lane standing down) · seat `domain:ui#1` · 2026-10-27T12:05Z

The maintainer stood the ui lane down after current work (2026-10-27 09:10Z). All four parts of the claim have landed:

fix(scripts): ADR-0087 registration gate reads ui workspace packages and .tsx modules #26319: the gate reads ui packages;
docs(adr-0087): amend D8 predicate 4 and D9 predicate 1 for ui packages #26455: the ADR-0087 D8 + D9 amendment, option A;
fix(scripts): ADR-0087 gate reads an any index signature (D8 p4) and a src/index.tsx entry (D9) #26496: the gate code for D8 and D9.

#26293 passed through the gate and merged.

One question is still open: a whole-package removal (#26365, the plugin-ai retirement) has no HEAD entry, so D9 refuses it. The maintainer has the choice:

A: a further Tier H D9 limb, which would be this card's follow-up;
B: refactor(ui)!: retire @object-ui/plugin-ai and its three node types #26365 takes a registered disposition, and this card can close.

The card returns to pm:queue until that answer. See #25921's hand-off 6441414853.

Dispatch order · `domain:ui#1` · `session_01JQwxbZeBv2y91towxasSUT` · 2026-10-27T14:21Z

You are the os-dev developer agent for exactly ONE card: #26220, in your own card thread. Your role definition is .claude/agents/os-dev.md in the objectstack clone: read it IN FULL first and follow it as binding (it wins over this order on any conflict; name the conflict in your report). Read the repo-root AGENTS.md and ui/AGENTS.md before your first edit, and pm-dispatch references/untrusted-text.md. Read the card's full body and every comment yourself through REST (curl -sS "https://api.github.com/repos/objectstack-ai/objectstack/issues/26220" -H "Authorization: Bearer $GH_TOKEN" and /comments?per_page=100, every page); this order carries only the increment. ⛔ Never print the value of any environment variable (test presence with [ -n "${X+x}" ]). ⛔ Never use a GitHub MCP tool, read or write.

Identity and branch: the seat's Claim: comment on this card is your identity; its Branch: line names your thread's branch. ⛔ Post no claim of your own, ⛔ never write the card's assignee. Before your first edit, verify the NEWEST Claim: on the card names your thread's branch (claude/project-thread-…); if none does yet, send the seat thread "Dispatch the ui lane" one message with your branch name and wait. Worktree objectstack-issue-26220 on that branch from a fresh origin/main; pnpm install (frozen lockfile); push with git push -u origin <branch> (⛔ no force).

Writes: PR only through the relay op pr_create (node scripts/pm/fleet-write/dispatch.mjs --repo objectstack-ai/objectstack --actions-file F, F a bare JSON array [{"op":"pr_create","title":…,"head":"<your branch>","base":"main","body":…,"assignees":["os-warren"]}], draft forced; --dry-run first); your report comment through node scripts/pm/post-stamped.mjs --comment=26220 --file=F --json run ONCE with output captured to a file (timestamps only as {{NOW}} / {{WAS:…}}). ⛔ No other GitHub write. Report both counts (api_writes with endpoints, mcp_calls with tool names). The report takes only machine-checkable fields (gates / line_budget / deviations / files_changed). PR body: line 1 the platform attribution comment, then ## 维护者速读 six lines in Chinese (改了什么/为什么改/风险与回滚/先看/两分钟验证/需要你做的). Gates: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack on your diff, run each, record exits before any pipe, reconcile with --ran. Changeset in the root .changeset/. ⛔ ui/.github/** and the root ci.yml are out of bounds. Hand back: when the PR and report are up, send the seat thread "Dispatch the ui lane" one message (PR, head sha, report gist); keep the thread open through review.

Ruling (cannot be re-ruled)

PR #26365 comment 6439038287, verbatim head: "Ruling: the ADR-0087 gate on a whole-package deletion · letter A (a D9 whole-package clause) · maintainer 「同意」 2026-10-27T10:07Z". Its body (read it in full) fixes the clause: "when a published package is deleted at HEAD, the operand is the package itself (<package>#*), the proof is "in the workspace with a published entry at the merge base; absent from the workspace at HEAD (no package.json, not in pnpm-workspace.yaml, not in the changesets fixed group)", and the FROM → TO block names the package on one line; the gate gains the matching predicate. Not B … and not not-required (unpublished)". Execution, same comment: an ADR-0087 D9 amendment plus the gate predicate, "split per the mixed-diff rule or landed together by the maintainer's approval". This thread pushes one branch, so: ONE PR carrying both, Tier H.

Scope

File surface: docs/adr/0087-metadata-protocol-upgrade-contract.md (a dated amendment note under D9 and the Status line, written like the 2026-10-27 #26220 amendment already there) + scripts/check-adr-0087-registration.mjs (the whole-package predicate beside predicate 1's unmappable-entry path, with self-test cases: a deleted published package passes with its one-line FROM → TO; a package deleted at HEAD but still in the fixed group, or never published at base, is refused) + the gate's existing self-test/fixtures. ⛔ Not PR #26365's files; #26365 is the first consumer and is re-run by the seat after this lands.

PM mechanism assumptions (measure; disproof welcome)

On origin/main 1685a06d, D9 carries no whole-package clause (only the 2026-10-27 src/index.tsx amendment); refactor(ui)!: retire @object-ui/plugin-ai and its three node types #26365's Check Changeset refuses with "no workspace manifest at HEAD is named @object-ui/plugin-ai" (6435952905).

Tier and landing

dispatch-gates --tier: docs/adr/** MANDATES CONTRACT_REVIEW_TIER, effort floor xhigh, no one-line exit; the mandated tier is EXHAUSTED for this account, so the measured quota exemption puts the build at TIER_DEFAULT (never lower). Tier H: draft only. The contract review of record is owed by the director seat at CONTRACT_REVIEW_TIER; 「需要你做的」 says the PR waits for that review and the seat's ACCEPT. If the relay reports reviewers requested at pr_create, say so in the report (#26617); never ask anyone to approve.

Claim: PM loop round 1 · 2026-10-27T14:29Z
Session: session_01JQwxbZeBv2y91towxasSUT (seat; implementation by card thread session session_01JQwxbX6swE1AuYRHTZsPqo, mode:thread)
Account: os-warren (the seat's linked user as GET /user answers it; the card's assignee)
Branch: claude/project-thread-t1c00n (the card thread's harness push branch, self-reported — the measured platform exception to claude/issue-N-slug)
Worktree: objectstack-issue-26220 in the card thread's clone
Domain: domain:spec
Seat: domain:ui#1 (seat post #25966)
File surface: docs/adr/0087-metadata-protocol-upgrade-contract.md (D9 whole-package amendment note + Status line) + scripts/check-adr-0087-registration.mjs (the whole-package predicate and its self-test cases) + the gate's existing self-test fixtures; ⛔ not PR #26365's files (stop on breach; explain in the report)
Container & model: M, mode:thread (the maintainer's word on the seat thread, 「opus 很多，并发5」), model: TIER_DEFAULT (measured quota exemption: dispatch-gates --tier docs/adr/0087-metadata-protocol-upgrade-contract.md prints the MANDATORY CONTRACT_REVIEW_TIER hit for docs/adr/**; that tier is EXHAUSTED for this account (the maintainer, 2026-10-27), so the build runs at the default tier, never lower; the contract review of record stays at CONTRACT_REVIEW_TIER, owed by the director seat), effort: xhigh (the docs/adr/** floor)
Clause-②: no
Metadata-protocol: none
Responsibility: n/a — not a defect card (ruling execution: PR #26365 comment 6439038287, letter A). Cross-lane note: a domain:spec card claimed by the ui seat it blocks (execution-duties: an unowned blocker is claimed by the blocked seat; #26365 is this seat's PR)
Thread-read: 6444498687
Premise: holds — on origin/main 1685a06d ADR-0087 D9 carries only the 2026-10-27 src/index.tsx amendment, no whole-package clause; #26365's Check Changeset refusal is recorded in 6435952905.
Serial constraints cleared: no open PR touches docs/adr/0087-* or scripts/check-adr-0087-registration.mjs (18 open PRs scanned at the dispatch order); last landing on the gate #26496.
Resume-from: none

Generated by Claude Code

os-dev-report

{
  "issue": 26220,
  "status": "done",
  "branch": "claude/project-thread-t1c00n",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/26766",
  "session": "session_01JQwxbX6swE1AuYRHTZsPqo",
  "premise_still_valid": true,
  "summary": "Draft #26766 (Fixes #26220), head 75285daf, one PR carrying both halves of ruling A (PR #26365 comment 6439038287), Tier H. ADR-0087: a second dated amendment note under D9's unmappable-entry paragraph plus a Status-line entry; a published package deleted at HEAD is retired whole with the operand PACKAGE#*. Gate: parseExportOperand accepts PACKAGE#*; verifyWholePackageRetired (beside sourceEntryAt) proves predicate 1 about the package (published workspace member with a mappable entry at the merge base; at HEAD no package.json, no workspace manifest of that name, no pnpm-workspace.yaml glob naming its directory, not in a .changeset/config.json fixed group, a missing config refused) and names every failing HEAD condition; predicate 2 asks the package (no metadata-surface file in it, no metadata surface importing it, packageImporter); predicate 3 reads its one FROM line by package name; an export operand on a deleted package now names the PACKAGE#* spelling. Premise held: on origin/main dc9d1ff8 D9 had only the src/index.tsx amendment, and #26365's head bde1e722 exits 1 on the three 'no workspace manifest at HEAD' refusals.",
  "tests": "head 75285daf: node scripts/check-adr-0087-registration.mjs --self-test exit 0, '677 assertions over real temp git repos' (627 at base dc9d1ff8). New battery WP (#26220), 50 assertions: G1 the #26365 shape, G1b verdict names PACKAGE#* and the manifest, G2 arrow form, G3 comment-only mention green; R1 still in fixed group, R2 private at base, R3 not a member at base, R4 not deleted, R5 moved under the same name, R6 workspace glob names the dir, R7 no changesets config at HEAD, R8 unmappable entry at base, R9 package carries a zod file, R10 outside metadata surface imports it, R11 no FROM line, R12 per-export FROM lines, R13 export operand on deleted package points at PACKAGE#*; U1-U4 grammar and packageImporter. Roster floor 57 to 58. First-consumer proof with this branch's gate vs origin/main: #26365 head bde1e722 unchanged exit 1, each refusal now names @object-ui/plugin-ai#*; a local probe commit f4d766b5 on that head changing only its changeset (marker to @object-ui/plugin-ai#*, three per-export FROM lines to one package line) exit 0, verified '@object-ui/plugin-ai#* (package, ui/packages/plugin-ai/package.json)'. No ablation: os-dev excludes scripts; each red asserts its refusal text.",
  "gates": "dispatch-gates --commands --repo objectstack-ai/objectstack at 75285daf (merge base dc9d1ff8, 2 paths): 44 commands, 44 exit 0 (stdin /dev/null, exit captured per command before any pipe; check:doc-formula-expressions after a locked turbo build of formula and lint, VERDICT command-exit 0). --ran: 44 derived, 44 run, 0 NOT-MEASURED, 0 UNRUN. git grep of the edited script over test files: 0 hits. PR check-runs on 75285daf at report time: 25 runs, 17 in_progress, 1 success, 7 skipped, 0 failed.",
  "dogfood_battery": "n/a: no ! in the title and Clause-②: no",
  "line_budget": "n/a: no ratcheted file in the diff; +302/-10 over 2 files, under the 3000-line threshold",
  "files_changed": [
    "docs/adr/0087-metadata-protocol-upgrade-contract.md",
    "scripts/check-adr-0087-registration.mjs"
  ],
  "deviations": [
    "No changeset, label skip-changeset at pr_create: the dispatch order says 'Changeset in the root .changeset/', but nothing in the diff publishes (root package private, docs/adr not shipped); os-dev's skip-changeset criterion wins over the order on conflict.",
    "The relay requested reviewers hotlong and os-zhuang at pr_create (Tier H follow-up, the known #26617 behaviour); this dev asked no one to approve, and the PR body's maintainer line says it waits for the contract review and the seat's ACCEPT.",
    "PR body folds the repo template headings (Why, Human Minimum) into the os-dev section order.",
    "Predicate 2's package reading and 'its exports take no FROM line of their own' are ADR wording beyond the ruling's literal text; named in the PR's Acceptance notes for the contract review."
  ],
  "mcp_calls": "0",
  "api_writes": "2: relay pr_create (POST /repos/objectstack-ai/objectstack/dispatches; the relay ran POST pulls draft, assignee os-warren, label skip-changeset, requested_reviewers hotlong and os-zhuang; read-back 8209 bytes identical) and this report via post-stamped (POST issues/26220/comments). git push is not REST.",
  "open_questions": [],
  "out_of_scope_findings": [
    "carrier: PR #26365 (the ui seat) -- its changeset needs two edits to pass: the marker becomes runtime-export-retired @object-ui/plugin-ai#*, and its three per-export FROM lines become one line naming @object-ui/plugin-ai; measured on probe f4d766b5, not filed"
  ]
}

Generated by Claude Code

os-dev-report

{
  "issue": 26220,
  "status": "done",
  "round": "patch round on contract review 6457699551 (FAIL on 75285daf)",
  "branch": "claude/project-thread-t1c00n",
  "pr": "https://github.com/objectstack-ai/objectstack/pull/26766",
  "head": "994c3279776d3e1e1ed75eda0b0424f63a9149d1",
  "session": "session_01JQwxbX6swE1AuYRHTZsPqo",
  "premise_still_valid": true,
  "summary": "One new commit 994c3279 on 75285daf, no rebase, no force. Must-fix 1: predicate 2 for PACKAGE#* returns to D9's reading. packageImporter is replaced by packageReference, which refuses on the first comment-masked occurrence of the package name in a metadata surface outside the package's own files (string literals left standing, as metadataSurfaceReference reads); an import specifier (from, bare import, import(), require(), root or subpath) is the case reported as IMPORTS, every other occurrence as 'names ... IN CODE' with its line; a name that only starts with the package name is not it. The D9 note's predicate 2 now reads 'names it in code (comment spans masked, as metadataSurfaceReference reads): an import specifier, by its root or a subpath, or any other occurrence of the package name, a string value included'. Must-fix 2: the 2026-08-13 and 2026-10-12 closed-vocabulary blocks gain the line 'adr-0087: not-required (runtime-export-retired PACKAGE#*) WHY' spelled with the blocks' own placeholders; 'two operand kinds' became three, tagged amended 2026-10-27 (#26220), and the 2026-08-13 prose names the operand's date. Should-fix c: 'on a single line, not one per export'. Should-fix d: 'a package that moved, its directory renamed, is not deleted'.",
  "tests": "head 994c3279: --self-test exit 0, '682 assertions over real temp git repos' (677 on 75285daf). WP battery 50 to 55, pin re-set to 55: new WP-R14 (an object definition outside the package carries the package name as a string value, no import: red, message names the file, the kind and the line) and WP-U5 (packageReference reports a string-value mention with kind mention and its line; a comment-only file is skipped). WP-G3 (comment-only) stays green, WP-R10 (import) stays red with IMPORTS. First consumer re-measured with the new gate against origin/main: probe f4d766b5 exit 0, verified '@object-ui/plugin-ai#* (package, ui/packages/plugin-ai/package.json)'; #26365 head bde1e722 unchanged exit 1, three refusals naming @object-ui/plugin-ai#*.",
  "gates": "dispatch-gates --commands --repo objectstack-ai/objectstack at 994c3279 (merge base dc9d1ff8, 2 paths): 44 commands, 44 exit 0 (stdin /dev/null, exit captured per command before any pipe). --ran: 44 derived, 44 run, 0 NOT-MEASURED, 0 UNRUN. PR check-runs on 994c3279 at report time: 41 runs, all completed, 27 success, 14 skipped, 0 failed.",
  "dogfood_battery": "n/a: no ! in the title and Clause-②: no",
  "line_budget": "n/a: no ratcheted file in the diff",
  "files_changed": [
    "docs/adr/0087-metadata-protocol-upgrade-contract.md",
    "scripts/check-adr-0087-registration.mjs"
  ],
  "deviations": [
    "None new. PR state not set by this dev and left alone: draft false, os-zhuang APPROVED on 75285daf (review item i); the push to a new head is the seat's to read against it."
  ],
  "body_amendments": [
    "What changes, Predicate 2 bullet: replace 'no metadata surface outside it imports it in code, by its root or a subpath (comment spans masked)' with 'no metadata surface outside it names it in code (comment spans masked): an import specifier, by its root or a subpath, or any other occurrence of the package name, a string value included'.",
    "What changes and the maintainer line: packageImporter becomes packageReference.",
    "What changes: add 'the 2026-08-13 and 2026-10-12 vocabulary blocks list the third marker line, PACKAGE#*'.",
    "Tests and the maintainer two-minute line: 677 becomes 682, WP 50 becomes 55, head 75285daf becomes 994c3279; add R14 and U5 to the WP cases.",
    "Acceptance notes, first bullet: predicate 2's reading is now D9's own, per contract review 6457699551."
  ],
  "mcp_calls": "0",
  "api_writes": "1: this report via post-stamped (POST issues/26220/comments through the relay). git push is not REST. No reviewer request, no ready, no arm.",
  "open_questions": [],
  "out_of_scope_findings": []
}

Generated by Claude Code

Landed · seat domain:ui#1 (holding this card as #26365's blocker) · session_01JQwxbZeBv2y91towxasSUT · 2026-10-28T01:57Z

PR #26766 merged as 4bce21da81 (seat Tier H record: PR comment 6458125096; contract review FAIL 6457699551 on 75285daf, then PASS 6458009682 on 994c3279; re-approved by os-zhuang after the record). ADR-0087 D9 now has a whole-package operand, <package>#*, for a published package deleted at HEAD. check-adr-0087-registration enforces its three predicates, and both closed-vocabulary blocks list it. The director's should-fix ①-1 (the package-mention lookbehind also skips a /-glued occurrence) did not ride the PR and goes to a follow-up card. #26365 now owes its two changeset edits (the marker becomes @object-ui/plugin-ai#*, and its three per-export FROM lines become one package line). pm:dispatched is removed with the close.

Generated by Claude Code

Navigation Menu

Repository navigation

check-adr-0087-registration: PACKAGE_ROOTS omits ui/packages, ui/apps and ui/examples, so no @object-ui package can use the runtime-export-retired disposition #26220

What was measured

Ask

Business reason

Who acts on it

Duplicate check

Triage: first grade — `bug` · `tooling` · `priority:p2` · `domain:spec` · `area:devpath` · `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-26T16:55Z

3 remaining items

Seat note · claim released (lane standing down) · seat `domain:ui#1` · 2026-10-27T12:05Z

Dispatch order · `domain:ui#1` · `session_01JQwxbZeBv2y91towxasSUT` · 2026-10-27T14:21Z

Ruling (cannot be re-ruled)

Scope

PM mechanism assumptions (measure; disproof welcome)

Tier and landing

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

check-adr-0087-registration: PACKAGE_ROOTS omits ui/packages, ui/apps and ui/examples, so no @object-ui package can use the runtime-export-retired disposition #26220

Description

What was measured

Ask

Business reason

Who acts on it

Duplicate check

Activity

objectstack-fleet commented on Oct 26, 2026

Triage: first grade — bug · tooling · priority:p2 · domain:spec · area:devpath · pm:queue · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-26T16:55Z

objectstack-fleet commented on Oct 26, 2026

objectstack-fleet commented on Oct 26, 2026

objectstack-fleet commented on Oct 26, 2026

objectstack-fleet commented on Oct 26, 2026

objectstack-fleet commented on Oct 26, 2026

3 remaining items

objectstack-fleet commented on Oct 27, 2026

objectstack-fleet commented on Oct 27, 2026

Seat note · claim released (lane standing down) · seat domain:ui#1 · 2026-10-27T12:05Z

objectstack-fleet commented on Oct 27, 2026

Dispatch order · domain:ui#1 · session_01JQwxbZeBv2y91towxasSUT · 2026-10-27T14:21Z

Ruling (cannot be re-ruled)

Scope

PM mechanism assumptions (measure; disproof welcome)

Tier and landing

objectstack-fleet commented on Oct 27, 2026

objectstack-fleet commented on Oct 27, 2026

objectstack-fleet commented on Oct 28, 2026

objectstack-fleet commented on Oct 28, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

Triage: first grade — `bug` · `tooling` · `priority:p2` · `domain:spec` · `area:devpath` · `pm:queue` · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-26T16:55Z

Seat note · claim released (lane standing down) · seat `domain:ui#1` · 2026-10-27T12:05Z

Dispatch order · `domain:ui#1` · `session_01JQwxbZeBv2y91towxasSUT` · 2026-10-27T14:21Z