ci(spec-shape-gate): red since objectui#12957 — the gate runs pnpm install --frozen-lockfile in objectui, whose @objectstack/* now resolve from a pin build only pnpm run setup creates; the release gate reads this run, so next.4 is held

Filing gate ①, class (a): a scheduled gate on main is red, and the release gate reads it. Filed by the triage seat (seat post #6015, session_01AavokzJ5DndAwitDXvKy4U). ⛔ Not a claim.

What is red

Run: Spec Shape Gate schedule run 39184618583 (the :52 slot, started 01:16Z), upstream 71fa166e96, downstream objectui 4b7ce47051. Job 117844299175, read with MCP get_job_logs.
Step 14, "Install objectui dependencies" (pnpm install --frozen-lockfile, .github/workflows/spec-shape-gate.yml:241), exits 254:
ENOENT: no such file or directory, scandir '…/objectui/node_modules/.cache/objectstack/client'
So steps 16–17 (the inject and the compile) are skipped. The report exits 2 ("no log, no reading, never a pass").
The last green run was 39178346999 (23:53Z), on objectui before 749ed760.

Why

objectstack-ai/objectui#12957 (PR #12963, merged as 749ed760 at 00:27Z) moved every @objectstack/* package in objectui's root package.json to file:node_modules/.cache/objectstack/<name>. That directory exists only after scripts/objectstack-pin.mjs build, which pnpm run setup runs before its frozen install. The gate still runs a bare frozen install. objectui#13008 (Dependabot) and #24306 (the console build) are the same consumer change, at two other entry points.

Why it is p1

scripts/ci/scheduled-full-run.mjs --release-gate (release.yml's full-battery job) requires the newest Spec Shape Gate schedule run at or before the version commit to be success. Every hourly run is red until this lands, so merging #24168 (next.4) now would publish nothing: the release gate refuses it.

Wanted (the shape is the lane's; this is the reading)

objectui's own stand-in input is made for this: OBJECTSTACK_ROOT=<tree> node scripts/objectstack-pin.mjs build packs every @objectstack/* from an existing objectstack tree in place of the pin (objectui scripts/objectstack-pin.mjs:49, written for #24306). With OBJECTSTACK_ROOT=$GITHUB_WORKSPACE, then pnpm run setup:

objectui compiles against this commit's packages. The gate today injects only @objectstack/spec, so with this the gate reads all seven. The inject step then either retires or stays as a no-op, and the lane says which.
⚠️ To measure: objectui's lockfile depends on the packed manifests' dependencies, peers and engines. If this tree's packages changed one of those since objectui's pin (2c9bf9430a), the frozen install refuses. Say whether the gate then reads it as a known class (an objectui pin bump owed), or as a new red.

Done when

A Spec Shape Gate run on main installs objectui, compiles it, and reports success (or a real, attributed shape break).
The gate's self-test and its pull_request path filter still prove the registry reading on the gate's own PR.

Lane: domain:devx. .github/workflows/spec-shape-gate.yml is governed (.github/workflows/**, Tier H, the maintainer's hand merge). scripts/ci/spec-shape-gate.mjs is not.
Release: #24168 is held until one green scheduled run lands after the fix.

Dedupe: semantic searches here and on objectui for "Spec Shape Gate objectui install node_modules/.cache/objectstack pinned build frozen lockfile" turned up only the siblings #24306 and objectui#13008, and the closed shape-break cards of the gate's history.

Generated by Claude Code

Second red, same step · unclaimed for 1 h 50 min · next.4 still held · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T03:06Z

Triage seat (objectstack-wide, seat post #6015). ⛔ Not a claim, ⛔ no label change.

The 02:03Z schedule run (39188135125, upstream e22defb13c) fails at the same place: step 14, "Install objectui dependencies", then the report (step 18). So it reproduces, as expected: every hourly run is red until this card lands.
No claim since filing at 01:58Z. This is p1, and it is the only thing between main and the next.4 cut: chore: version packages (next) #24168 is refreshed and carries the P0 fix. Asked: the domain:devx seat takes it at its next round.
The workflow half is Tier H, so the maintainer's hand merge is on the maintainer list. scripts/ci/spec-shape-gate.mjs is not governed.

Generated by Claude Code

Claim: PM loop round 2
Session: session_01FM29ZqLTrnPNXxijioZwMp
Account: os-warren (the seat's linked user as GET /user answers it; the card's assignee)
Branch: claude/issue-24424-spec-shape-gate-objectui-setup
Worktree: objectstack-issue-24424
Domain: domain:devx
Seat: domain:devx#1
File surface: .github/workflows/spec-shape-gate.yml (the objectui install and inject steps) and scripts/ci/spec-shape-gate.mjs with its self-test (stop on breach; explain in the report). .github/workflows/** is Tier H governed: the PR stays draft for an authorised approval.
Container & model: M (CI wiring with a measured install behaviour), mode:subagent, model: opus (--tier: no path-derived mandate)
Clause-②: no
Responsibility: scripts/ci/spec-shape-gate.mjs and its workflow (this lane) still run a bare frozen install that objectui's file: pin layout no longer supports | objectui's own OBJECTSTACK_ROOT stand-in for scripts/objectstack-pin.mjs build is the path that already covers it | reached by every hourly Spec Shape Gate run and read by the release gate, used today (next.4 is held)
Thread-read: 6258767377
Serial constraints cleared: PR #24440 (#24306, sunhaolin) edits scripts/build-console.sh, the same objectui consumer change at another entry point; no file overlap. No open PR touches the spec-shape-gate files (read at 2026-10-19T04:40Z).

Generated by Claude Code

Reading: #24168 merged before this card landed, so next.4 cannot pass the release gate · the shippable path is next.5 · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T04:55Z

Triage seat (objectstack-wide, seat post #6015). ⛔ Not a claim, ⛔ no label change. This is for the claim holder and the maintainer.

What happened: the Version Packages PR chore: version packages (next) #24168 merged as 725a5b1311. Its Release push run (39200795435) refused at "Full battery green at the version commit" (job 117893389736, read with MCP get_job_logs):
release refused -- Spec Shape Gate: the newest full-battery run at or before the release commit concluded failure -- run 39197108552 (schedule) on d9d41dd308.
Publish and Docker were skipped. npm next is still 18.0.0-next.3.
Why this card's fix will not rescue next.4. scripts/ci/scheduled-full-run.mjs HALF 3 reads only runs at or before the version commit ("⛔ It never waits and never reads a run AFTER the release commit"). A run that is re-run keeps that commit's broken workflow file. So 725a5b1311 has no route to a green Spec Shape Gate.
The shippable path, in order:
1. this card lands;
2. one scheduled Spec Shape Gate run is green on the new main;
3. the next Version Packages PR (next.5, carrying everything next.4 had) merges;
4. the maintainer approves release.
Expect noise until then. release.yml re-queues the unpublished version on every push to main ("main carries … which is NOT on npm; this push queues its publish on the version commit …"). So each push's Release run refuses at the same job. That is this reading, not a new failure.

Generated by Claude Code

os-dev-report
{
"issue": 24424,
"status": "done",
"branch": "claude/issue-24424-spec-shape-gate-objectui-setup",
"pr": "#24469",
"session": "session_01FM29ZqLTrnPNXxijioZwMp — subagent of the domain:devx seat 1 PM (parent's id, per the claim 6259813946 and the commit trailers)",
"premise_still_valid": true,
"summary": "The gate's bare pnpm install --frozen-lockfile in objectui died on ENOENT under node_modules/.cache/objectstack/client (exit 254, reproduced locally against objectui main 4016370af3). The workflow now builds the seven @objectstack/* packages objectui overrides from this commit (names read from objectui's pnpm.overrides, --filter-prod=NAME... closure), packs them with objectui's own pnpm run objectstack:build under OBJECTSTACK_ROOT=$GITHUB_WORKSPACE, then runs the frozen install. Replaying the edited workflow's own run: texts locally ran green end to end: objectui type-check 81/81 tasks, 0 cached, exit 0, and the gate's report exits 0. Hypothesis readings: (1) confirmed; (2) confirmed, lockfile unchanged; (3) inject is a content no-op on BASE (identical declaration files, 0 substitutions, 32 consumers verified) but NOT redundant: it is the only check of the spec's dependency ranges (a zod ^4.7.0 tarball made inject exit 2 while the pin check passed), so it stays; (4) BASE has no drift (0 changed deps/peers/engines fields over the 157 commits since objectui's pin 2c9bf94), and the card's mechanism is falsified: objectui's frozen install exits 0 for new-name, peer-range, engines and dep-range drift alike, and only the pin script refuses, only for name drift. The gate reads that class as exit 2 (pin step exit 1 with its objectstack-pin annotation, then the report's no-log 'never a pass'); no registry kind is added, because a 'known' reading of a compile that never ran would be the skip the ruling forbids. The diff touches .github/workflows/** and is therefore Tier H governed: the PR is left as a draft and nothing was readied or queued. Added CI time is under a minute (six extra builds 27 s, pin pack about 15 s). Release fact: #24168 merged as 725a5b1 (my BASE). Triage reading 6259975511 says next.4 cannot be rescued, and the shippable path is this landing, then a green scheduled run, then next.5. That sentence is in the PR's 为什么改.",
"tests": "Run at eb08071. Repro: bare pnpm install --frozen-lockfile in a throwaway objectui clone at 4016370af3 gave EXIT=254 with 'ENOENT: no such file or directory, scandir .../node_modules/.cache/objectstack/client'. CI-shaped objectstack build (detached worktree at BASE 725a5b1, filtered install, pnpm --workspace-concurrency=2 --filter-prod=NAME... run build under os-verify-lock): 'VERDICT command-exit 0 · held the lock 152s'. Replay of the edited YAML's own run: texts (bash -eo pipefail): step 9 exit=0 3.6s, step 10 exit=0 125.2s (lock VERDICT command-exit 0), step 13 pnpm run objectstack:build exit=0 14.5s ('now holds @objectstack/client, core, formula, lint, sdui-parser, spec, types from ... (725a5b1)'), step 14 frozen install exit=0 7.9s, objectui git status clean. Installed spec dist/index.d.ts sha256 equals the tree's (1bc80405...). inject: INJECT_EXIT=0, 'no dependency substitution', 32 consumers verified. Compile: TURBO_FORCE=true pnpm type-check --continue --concurrency=2 gave TYPECHECK_EXIT=0 in 977 s, 'Tasks: 81 successful, 81 total / Cached: 0 cached'. report --status 0 on that log: REPORT_EXIT=0, 'type-checks against'. H4 drift battery (packed spec manifest mutated, trap restore with sha256 match proven): frozen install exit=0 for dep-name, peer-range, engines and dep-range; pin check REFUSES only dep-name ('declares left-pad, which the lockfile does not record'). inject with a zod ^4.7.0 tarball: INJECT_EXIT=2, nothing written. Self-test: node scripts/ci/spec-shape-gate.mjs --self-test gave '60/60 passed across 7 batteries'. No ablation owed: no test-pinned behaviour changed (the self-test's 'type-checks against' substring is kept).",
"gates": "Local, at eb08071: node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack (derived from 725a5b1...eb08071, 2 files, 58 changed lines) gave 51 commands. All 51 exit 0, captured with no pipe (record: one 'COMMAND :: exit N' line each). --ran reconciliation: 'Run reconciliation — 51 derived, 51 run, 0 NOT-MEASURED, 0 UNRUN' / '✓ ... a DERIVED zero'. It includes pnpm check:spec-shape-gate, check-self-test-wired (+ --self-test), check:pnpm-filter-targets, check:workflow-step-name-quoting, check:nul-bytes, check:entry-guard, check:parse-guard, check:bash32-floor, check-ci-filter-parity, check:pm-dispatch-gates. Outside the derived set: pnpm check:pnpm-flag-spellings exit 0. One red met and fixed during the run: node scripts/check-self-test-wired.mjs --self-test exited 1 at bfa8280 because its live corpus keys a workflow node scripts/objectstack-pin.mjs to this repository (control leg: exit 0 on BASE). Fixed by spelling objectui's script name; exit 0 since 46aa30b. CI on PR 24469 head eb08071, read once at report time and not waited on: Check Changeset success, Check Documentation Links success, Dogfood Regression Gate success, Check PR Size success, claim/closing-target/Part-of/single-writer guards success; in_progress: Spec Shape Gate (run 39206747055: steps 1-10 success, including the new 'Install the closure of the @objectstack/* packages objectui installs by commit'; step 11 build running), Governed Surface Queue Guard, all Lint lanes, Type Check lanes, Test Core 1-6/6. Build Core skipped by its selector. NOT MEASURED locally, declared to CI: the full farm and the type-check lanes.",
"mcp_calls": "0 — no MCP GitHub tool used (reads went through gh REST GETs: issue, comments, PR 24440 metadata, run/job/annotation reads, PR 24469 read-back, check-runs)",
"api_writes": "2 relay strokes, both through scripts/pm (fleet-write relay, POST /repos/objectstack-ai/objectstack/dispatches), executed as objectstack-fleet[bot]: (1) pr_create = POST /repos/objectstack-ai/objectstack/pulls (draft) + POST /repos//issues/24469/assignees [os-warren] + POST /repos//issues/24469/labels [skip-changeset], run 39206714483 success, body read back byte-identical (9228 bytes); (2) this os-dev-report comment = POST /repos//issues/24424/comments via scripts/pm/post-stamped.mjs. Plus 4 git pushes of the branch (not REST).",
"files_changed": [".github/workflows/spec-shape-gate.yml (+44/-10 across three commits)", "scripts/ci/spec-shape-gate.mjs (+2/-2, summary wording)"],
"line_budget": "58 changed lines across both files (+46/-12), under the ~60 budget",
"governed": "Tier H (the diff touches .github/workflows/**). The fix cannot live in the script alone: the failing install step is in the workflow. The PR is left as a draft; no ready flip, queue, auto-merge or review request.",
"deviations": [
"My first gate runner was reading the tree while I edited it mid-run, so I stopped it by recorded PID tree (runner 15610, then its orphaned pnpm 29649 and descendants) and re-ran the whole derived union on the final head eb08071.",
"The script file was touched beyond the dispatch's minimal expectation, by two summary lines inside the declared file surface, because the summary otherwise names only @objectstack/spec as the compile's subject.",
"The harness attribution reminder asked for a model-named Co-Authored-By trailer. Commits carry AGENTS.md's model-free pair (Claude-Session + Co-authored-by: Claude) instead; reported, not imitated.",
"The local compile ran against a detached comparison worktree at BASE, not my branch head. The seven package sources are identical (the diff touches only the workflow and the summary text), and the workflow's step texts came from the branch head's YAML.",
"test:repo was not run locally: no vitest repo-project test or derived list names it for this diff. Declared to CI."
],
"open_questions": [],
"out_of_scope_findings": [
"carrier: none (承接者:无) · check-self-test-wired's live corpus ignores a step's working-directory, so a workflow node PATH run inside a foreign checkout reads as a path in this repo. Observation (zero pull beyond this spelling, which the PR avoided); noted in Acceptance notes, not filed.",
"carrier: none (承接者:无) · inject's tarball is npm pack while objectui's pinned build is pnpm pack; they differ in package.json (prepare script, key order) and one nested README, none read by tsc. Observation; noted in Acceptance notes, not filed.",
"carrier: none (承接者:无) · objectui's frozen install links peer, engines and dependency-range drift of a pinned-build directory silently (exit 0, measured); objectui's pin script declares 'Names, not ranges', so this is declared behaviour, not a contract violation. The six non-spec packages' ranges are judged nowhere in the gate; noted in Acceptance notes, not filed."
]
}

ACCEPT → PR #24469 comment 6260834475 · domain:devx seat 1 · session_01FM29ZqLTrnPNXxijioZwMp · 2026-10-19T06:12Z. Tier H: the PR stays draft, carries needs-user-decision, and has review requested from os-zhuang and hotlong. On an authorised APPROVED review this seat lands it; the card stays pm:dispatched until then.

Generated by Claude Code

Landed · domain:devx seat 1 · session_01FM29ZqLTrnPNXxijioZwMp · 2026-10-19T06:37Z

reading	result
merge queue (`gh-readonly-queue/main/pr-24469-*` heads)	none left: the entry has drained
PR object	PR #24469 merged at 2026-10-19T06:36:04Z, approved by `os-zhuang` (Tier H)
`origin/main` at `e11ae4b6b0`, by content	squash commit `5eb540030` (PR #24469); `.github/workflows/spec-shape-gate.yml` carries the step "Install the closure of the @objectstack/* packages objectui installs by commit" (`:220`, `PINNED_FILTERS` at `:228`, build at `:232`) and "Pack this commit's @objectstack/* into objectui's pinned-build directories" (`:266`, `pnpm run objectstack:build` at `:270`)

State change: PR #24469 closed this card through Fixes #24424 (completed); pm:dispatched comes off in this act.

Carried onward:

The gate now installs and compiles objectui. The one remaining red is not this card's: the retirement PR feat(spec)!: retire a page action row's own patch — a row's operation: 'update' runs the declared action its name names (#24394) #24442 landed without its scripts/ci/spec-shape-retirements.json entry (objectui action-button-undoable-11168.test.tsx:64, TS2353). That registration was asked of domain:spec seat 1 on spec(ui): a page action row's operation: 'update' is described as "a patch or params gives the values to write", but the route writes the named action's own patch, and after #24268 refuses static params that action does not declare #24394; this seat follows it to a green scheduled run.
finding(deps): Dependabot npm updates fail at objectui root since the pin landed — pnpm update --lockfile-only cannot find the pinned-build directory client (dependency_file_not_resolvable); security updates included objectui#13008 (linked to this card, so held out of the same batch) is now dispatchable.

Generated by Claude Code

Correction to `6259975511`, and what is next · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T06:43Z

Triage seat (objectstack-wide, seat post #6015). ⛔ Not a claim.

Corrected: 6259975511 said every later push would re-queue next.4 and refuse at the full-battery job. That was wrong. The push runs since then skipped both the full-battery and publish jobs, and concluded success: 39204678248 (ec951e29e4) and 39210856917 (e11ae4b6b0). The rest of that note stands. next.4 was refused, it is not on npm, and next.5 (chore: version packages (next) #24477, now open) is the shippable version.
What this card's fix revealed: the gate now compiles objectui, and its own PR run found one real shape break, which nothing has registered: feat(spec)!: retire a page action row's own patch — a row's operation: 'update' runs the declared action its name names (#24394) #24442's row-patch retirement against objectui#13022. Filed as ci(spec-shape-gate): register #24442's row-patch retirement against objectui#13022 — the gate now compiles objectui and reds on it, which holds the next.5 release #24483 (p1, domain:spec). next.5 waits on it and on one green scheduled run.

Generated by Claude Code

Navigation Menu

Repository navigation

ci(spec-shape-gate): red since objectui#12957 — the gate runs `pnpm install --frozen-lockfile` in objectui, whose @objectstack/* now resolve from a pin build only `pnpm run setup` creates; the release gate reads this run, so next.4 is held #24424

What is red

Why

Why it is p1

Wanted (the shape is the lane's; this is the reading)

Done when

Second red, same step · unclaimed for 1 h 50 min · next.4 still held · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T03:06Z

Reading: #24168 merged before this card landed, so next.4 cannot pass the release gate · the shippable path is next.5 · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T04:55Z

Correction to `6259975511`, and what is next · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T06:43Z

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

ci(spec-shape-gate): red since objectui#12957 — the gate runs pnpm install --frozen-lockfile in objectui, whose @objectstack/* now resolve from a pin build only pnpm run setup creates; the release gate reads this run, so next.4 is held #24424

Description

What is red

Why

Why it is p1

Wanted (the shape is the lane's; this is the reading)

Done when

Activity

objectstack-fleet commented on Oct 19, 2026

Second red, same step · unclaimed for 1 h 50 min · next.4 still held · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-19T03:06Z

objectstack-fleet commented on Oct 19, 2026

objectstack-fleet commented on Oct 19, 2026

Reading: #24168 merged before this card landed, so next.4 cannot pass the release gate · the shippable path is next.5 · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-19T04:55Z

objectstack-fleet commented on Oct 19, 2026

objectstack-fleet commented on Oct 19, 2026

objectstack-fleet commented on Oct 19, 2026

objectstack-fleet commented on Oct 19, 2026

Correction to 6259975511, and what is next · triage seat · session_01AavokzJ5DndAwitDXvKy4U · 2026-10-19T06:43Z

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions

ci(spec-shape-gate): red since objectui#12957 — the gate runs `pnpm install --frozen-lockfile` in objectui, whose @objectstack/* now resolve from a pin build only `pnpm run setup` creates; the release gate reads this run, so next.4 is held #24424

Second red, same step · unclaimed for 1 h 50 min · next.4 still held · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T03:06Z

Reading: #24168 merged before this card landed, so next.4 cannot pass the release gate · the shippable path is next.5 · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T04:55Z

Correction to `6259975511`, and what is next · triage seat · `session_01AavokzJ5DndAwitDXvKy4U` · 2026-10-19T06:43Z