[objectstack-fleet]

Requested by Tesla · project thread

Part of #25712 (4b PR 2d-2a: the per-gate wiring pins among the ui/scripts/__tests__ files that scan ui/.github/workflows/ stop reading it).
Part of objectstack#25677.

Clause-②: no
Metadata-protocol: none

维护者速读(草稿)

• 改了什么: ui/scripts/__tests__ 里有 18 个测试文件扫描嵌套的 ui/.github/workflows/ 目录,问的是“哪个 workflow 运行这个 gate”。这个目录在 monorepo 里根本不运行。其中 14 个文件的 gate 唯一的家是 objectui 的 repo-gates.yml,根目录没有任何 workflow 运行它们。它们的接线 pin(home、trigger、path filter、install 顺序、step 名、header 计数)退役,注释写明由 ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880 承接(ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880 把这 19 个 gate 移植进根 Lint UI,届时把 pin 指回那个 job)。不读 workflow 的用例全部保留。另外 4 个文件断言“没有 workflow 运行这个 gate”,改为扫描根目录 .github/workflows/(新的 rootWorkflowFiles())。
• 为什么改: 2d-3 要删除 ui/.github/workflows/。删之前这 18 个文件必须先不依赖它。改完以后,不管 ui/.github/workflows/ 在不在,这 18 个文件都是绿的(本地两种布局都是 1,034 个用例通过)。
• 风险与代价(含回滚): 只动测试、一个测试 helper 和一个测试 ledger,不动任何会运行的代码、workflow 或受管面。退役的 pin 有 ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880 接回。回滚 = revert 本 PR。
• 席位意见:(留空)
• 你要做的: 无。Tier S,由席位落地。

席位核对

(left for the dispatch seat)

Why

GitHub runs only the repository root's .github/workflows/. Since the cutover (#25825), objectui's workflows sit at ui/.github/workflows/ and run nowhere. 2d-1 (#25879) moved the 28 files that read ui/.github/workflows/ci.yml to the root ci.yml. This PR takes 18 of the 27 files that scan the workflow directory as a population. The dispatcher split the population in two: the other 8 (the governance files, which read through workflow-checks.ts or scan every workflow for a repo-wide rule) are 2d-2b. check-pre-install-import-graph stays with 2d-3, because its gate, not just its test, derives its population from the nested workflows.

The rule is the 2d-1 rule: retarget a pin to the root workflows by path, or retire it with a comment naming who carries its subject. Nothing is skipped and nothing is folded into a root workflow.

The 19 gates in ui/.github/workflows/repo-gates.yml run in no root workflow. #25880 ports them into the root Lint UI job. Per the dispatcher, the pins on their wiring retire against #25880, not against roster row 24.

What changes

• 14 files: gate-wiring pins for repo-gates.yml gates retire against ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880. Each file loses its ui/.github/workflows reads and the cases built on them: has a workflow that gates pull requests, runs it in NO path-filtered workflow, has exactly one home, the install-order half of needs no install, subscribes merge_group, step-name assertions, does not run a NEIGHBOUR gate, and the header-count pins. The note where they were names ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880. Every case that reads no workflow stays:

  • the package.json aliases;
  • the import-graph walks of needs no install (check-doc-component-types, check-prompt-component-keys), kept as needs no install: its import graph reaches no package;
  • check-doc-snippet-types's build-filter emission cases;
  • the REQUIRED_CONTEXTS and dependabot-merge-gate.mjs classification halves;
  • the CI-page section and by-command row pins;
  • check-doc-component-types's three header-count controls, which exercise documentCounts and headerComments on synthetic text. They become the controls of the pin ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880 brings back.

• check-doc-snippet-types' the build-filter steps propagate failure… describe (6 cases) retires. It ran the run: scripts of two repo-gates.yml steps under bash. ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880 carries those steps, and the cases come back against them.

• 4 files: "no workflow runs this" pins ask the root directory. check-hand-rolled-comment-mask, check-i18n-dead-keys, check-lint-rule-coverage and check-published-dist-tooling (the per-PR half) each asserted that no workflow runs a gate. They now read the root .github/workflows/ through helpers/root-ci.ts:

  • rootWorkflowsDir;
  • rootWorkflowFiles(), which refuses a scan that reads fewer than 10 files, so a "nothing names this" pin cannot go vacuous;
  • rootWorkflow(file).

  check-published-dist-tooling's runs nightly read published-dist-gate.yml and retires as roster row 23 (kept, follow-up port).

• ui/scripts/markdown-test-inputs.mjs. Its scanner flagged every removed readdirSync(workflowDir) as a directory walk, so the ledger changes:

  • ten entries drop that walker: declaration;
  • three entries the scanner no longer offers at all are removed: check-i18n-dead-keys and check-lint-rule-coverage (both reads: []), and check-prompt-component-keys.

  The prompt-keys test still reads .github/prompts/*.prompt.md, but through the gate's exported PROMPT_DIR, which the scanner cannot see. 2d-3 moves those prompts to the root .github/prompts/ (dispatcher ruling), outside the tree this ledger describes. Test UI runs the whole suite on every ui change, so no selection depends on the entry today.

Disposition table (18 files)

#	File	Retired (reads repo-gates.yml unless noted)	Kept	Carried by now
1	check-doc-component-types	workflow, path filter, one home, install order, header-count pin	alias, import-graph walk, 3 header-count controls	#25880
2	check-doc-example-ids	workflow, path filter, one home, install order, header count	alias, neighbour-alias pin	#25880
3	check-doc-example-types	workflow, path filter, one home, after build, after sibling; the alias case's workflow half	CI-page by-command row, alias	#25880
4	check-doc-json-fences	one home + filter, after build and siblings	alias + CI page	#25880
5	check-doc-snippet-types	workflow, path filter, one home, builds a filter, builds before; the objectui#6221 step-script describe (6)	build-filter emission (3), alias	#25880
6	check-prompt-component-keys	one home, unfiltered, install order	alias, import-graph walk	#25880
7	check-readme-exports	one home, no trigger filter, triggers, builds before; step names	alias, REQUIRED_CONTEXTS classification	#25880
8	check-shell-escape-residue	workflow, merge_group, path filter, one home, install order	alias, REQUIRED_CONTEXTS classification	#25880
9	check-skill-eval-tokens	workflow, self-test, installs nothing, path filter, one home, neighbour; step name	alias, classification, CI-page heading	#25880
10	check-skill-examples	workflow, self-test, derived build filter, path filter, one home, neighbour; step name	alias, classification, CI-page heading	#25880
11	check-skills-paths	workflow, path filter, one home	alias, baseline	#25880
12	check-vi-mock-inherit	workflow, merge_group, path filter, one home, install order	alias	#25880
13	check-vi-mock-specifiers	workflow, merge_group, path filter, one home, install order	alias	#25880
14	docs-links-workflow	runs the checker, triggers, no path filter, one home, docs-only startable; the workflow half of exists at all	script exists, alias	#25880
15	check-hand-rolled-comment-mask	(none)	nothing GitHub runs names this gate	retargeted: root .github/workflows/
16	check-i18n-dead-keys	(none)	is NOT invoked by any GitHub workflow	retargeted: root .github/workflows/
17	check-lint-rule-coverage	(none)	the inverse pin, not in a workflow yet	retargeted: root .github/workflows/
18	check-published-dist-tooling	runs nightly (published-dist-gate.yml)	is NOT wired into any per-PR workflow	per-PR half retargeted: root .github/workflows/; nightly half roster row 23

Some pins outlive their meaning but still read no workflow. For example, the CI page still says the doc gates block. They stay green and are #25880's to restate when the gates run again.

What 2d-2b and 2d-3 still owe (not here)

• 2d-2b (8 files): dependabot-merge-gate, merge-queue-reporting, check-lockfile-dedupe, check-lockfile-integrity (all through workflow-checks.ts), ci-setup-pnpm-wiring, workflow-cache-save-bound, doc-version-claims, and labeler-package-coverage. The last of these is retargeted to the root .github/labeler.yml, which gains the ui/ globs it lacks (dispatcher ruling on F12).
• 2d-3: check-pre-install-import-graph, check-action-ref-convention (its gate scans the nested workflow directory), and the 24 single-workflow readers. It also moves the three ui/.github/prompts/*.prompt.md files to root .github/prompts/, with a separate prompts root for check:prompt-keys, and deletes ui/.github/dependabot.yml and the workflows once ci(ui-merge): the 19 objectui repo gates in ui/.github/workflows/repo-gates.yml have run nowhere since the cutover — port them into the root Lint UI job before 2d-3 deletes the nested workflows (roster row 24; parent #25677) #25880 lands.

Verification

Run from ui/ per ui/AGENTS.md, at this head (main 6e0597f2 plus this commit):

Run	Result
The 18 modified test files, ui/.github/workflows/ present	18 files, 1,034 tests, all passed
The same 18 files with ui/.github/workflows/ moved aside (the 2d-3 layout)	18 files, 1,034 tests, all passed
The same 18 with all of ui/.github/ moved aside	17 pass; check-prompt-component-keys fails 7 prompt-surface cases, because the prompts are under ui/.github/prompts/. 2d-3 moves them (above).
Whole scripts/__tests__ suite, --shard=1/4 through 4/4, under scripts/pm/os-verify-lock.sh	5,403 passed, 1 failed, 2 skipped, after the ledger fix
Every test that imports markdown-test-inputs.mjs (10 files)	340 passed
pnpm type-check:scripts	the one error is packages/components/src/renderers/action/resolve-icon.ts(265,69) TS2339, identical on unmodified main here; none in a touched file
pnpm exec eslint --max-warnings=0 (touched files), lint:coverage, check:test-path-roots, check:vi-mock-override-shape; root pnpm check:test-source-alias	all exit 0

The one suite failure is the base red riding main since 2c: check-pre-install-import-graph, unresolved: scripts/objectstack-pin.mjs. It clears in 2d-3 with the nested workflows.

Derived gates (node scripts/pm/dispatch-gates.mjs --commands --repo objectstack-ai/objectstack from this diff): 18 families, all exit 0 at this head. --ran reconciles 18 run, 0 NOT-MEASURED, 0 UNRUN. The slowest were check:query-options-erasure (320 s), check-comment-mask-corpus (202 s) and check:engine-double-contract (68 s).

The push-environment battery (GITHUB_ACTIONS=true GITHUB_EVENT_NAME=push node scripts/pm/check-dispatch-gates.mjs, both tiers) is running on this head. Its line goes to the dispatch seat.

🤖 Generated with Claude Code

https://claude.ai/code/session_01A6SEitNngAMugGbuJHWsDS

[objectstack-fleet]

Standing down on two red Test UI shards at 8086728e. Neither is this PR's.

• Test UI (11/16) fails ui/scripts/__tests__/check-pre-install-import-graph.test.ts:297 with unresolved: scripts/objectstack-pin.mjs. This is main's own red: main's push run 40264138159 (cb0a2fcf) fails the same shard on the same row. Nested workflows under ui/.github/workflows/ still call node scripts/objectstack-pin.mjs before install, and release(ui-merge): one release lane for both scopes — @object-ui joins the fixed group, @objectstack/console builds from ui/apps/console, the objectui pin is deleted (#25711) #25745 deleted that script. No fix exists yet. tooling(ui-merge): governance in-repo — liveness evidence and spec tests stop citing objectui by commit, one sdui-parser, AGENTS.md / pm scripts drop objectui#, ui/AGENTS.md nests, 18 governance workflows merged or retired (parent #25677, 4b/6) #25712's PR 2d-3 removes those workflows and retargets the gate's population to the root workflows. The failed-jobs re-run reproduced it identically (job 121388332191).
• Test UI (10/16) failed ui/apps/console/src/pages/settings/__tests__/SettingsView.shellLeave-13231.test.tsx (after a successful save: the shell leaves at once). This PR does not touch that file or anything it imports. The file passed 3 of 3 times locally at this head, and the one failed-jobs re-run turned the shard green (job 121388332209).

Neither shard is a required context.

---

Generated by Claude Code

[github-actions]

⛔ merge queue 构建失败 — 先分诊，再决定要不要重排

队列构建 40267487698 红了。队列跑的是全量套件（PR 侧 CI 只跑 affected 子集），
所以失败的测试可能在本 PR 没碰过的包里 —— 那不是重排能修的。每次盲目重排都会让排在后面的所有 PR 重建一轮。

分类：failure —— 按下面的日志分诊。

失败的 job（日志抽取，best effort）：

• Test UI (13/16) — 失败步骤: Run the ui tests (shard 13/16)

  × finds no pre-install gate reaching a package 9ms
  FAIL   unit  scripts/__tests__/check-pre-install-import-graph.test.ts > the real tree — every step objectui#6148 measured is still derived > finds no pre-install gate reaching a package
    ↳ 失败原因: AssertionError: a script this repository runs before `pnpm install` needs `node_modules` to load: expected [ Array(1) ] to deeply equal []
  

↳ 失败原因 是判读的关键：超时（Test timed out in … / Hook timed out in …）多半是负载/时序，不是本 PR 的回归；
断言（AssertionError: …）才指向真实的行为改变。两者的 FAIL 行长得一模一样，只有这一行能区分。

⚠️ 断言这一侧有一类例外，判据是断言在测什么，不是它是不是 AssertionError。 断言的对象是产品行为（一个值、一个形状、一次拒收）⇒ 照上面读：真实的行为改变，去查，⛔ 不要重排掉；
断言的对象是这次实验自身的有效性前提（跑完的耗时、负载下的先后、任何只在时间预算内才成立的条件）⇒ 它跟超时是同一类，同样对负载敏感，重排一次是合法的判别手段。
识别是机械的：断言的消息或它比较的值本身点名了一段时长、一个时间戳、一个耗时计数。实测过的一对 —— AssertionError: SecurityPlugin.init() ran: expected false to be true 测的是产品行为（真回归）；
AssertionError: this run took over a second, so second-precision stamps could have differed too: expected 1006 to be less than 1000 测的是实验前提：它守护的那条不变式当时是绿的，同一个 head 原样重排一次即成功。
穿着 AssertionError 外衣的时间测量，仍然是时间测量。（⛔ 这只改「怎么读一次红」，不改「哪些测试可以重排」——后者由别处管。）

跨 PR 相同签名（24h，按失败测试文件聚合）：

• scripts/__tests__/check-pre-install-import-graph.test.ts — 24h 内已弹出 11 个 PR（feat(spec,cli): reports are translatable — reports in the translation bundle #25826、fix(ui-merge): two tests that #25745 broke on main read the merged tree — the Changesets config through changesetConfigPath(), the SDUI html tier by its tags (#25711) #25852、ci(ui-merge): Type Check UI builds the five vitest-setup packages before that step, so an empty ui scope no longer reds it #25856、ci(ui-merge): register Lint UI and Build & E2E UI in REQUIRED_CONTEXTS #25857、fix(ui-merge): the CI page and two of its tests stop reading the two workflows #25745 deleted #25861、test(ui): three ui/ script tests spell their temp bases so the whole-tree scratch-dir sweep reads them (PM dispatch-gates red on main) #25865、ci(ui-merge): register Type Check UI in REQUIRED_CONTEXTS #25878、test(ui-merge): the ui gate-wiring pins read the root ci.yml that runs them; pins on jobs it did not port retire #25879、ci(ui-merge): queue builds decide the four ui lanes from the merged diff against merge_group.base_sha #25881、ci(ui-merge): Test UI runs only the ui tests a pull request or merge group affects — the root ci.yml wires ui/scripts/select-affected-tests.mjs #25882、test(ui-merge): the repo-gates wiring pins retire with their port; "no workflow runs this" pins read the root workflows #25904），分属 2 条投机栈 ⇒ 2 次独立命中。汇总 issue：Queue-flake anchor: scripts/__tests__/check-pre-install-import-graph.test.ts #25863（已刷新）
• ⚠️ 24h 评论账本没读完（超过 5 页仍未读到窗口尽头），所以上面的「不同 PR 数」是下界，不是全量。

历史信号：

• 本 PR 过去 24h 无队列失败记录（首次）。
• 过去 24h 队列共有 13 个失败构建（不含本次）。

分诊清单：

1. 失败测试在本 PR 改动的包里 → 真回归，修 PR。
2. 失败测试与本 PR 无关 → 看上面的「跨 PR 相同签名」；已有汇总 issue ⇒ flaky/环境问题实锤，去那张 issue 上谈，修好前重排只会再烧一轮全队列。
3. 两者都不是 → 可能与同组 PR 语义冲突；等前面的 PR 落地或失败出队后再重排一次即可，不要连续重排。

---

Generated by Claude Code · merge-queue-triage workflow (#4859)

[github-actions]

⛔ merge queue 构建失败 — 先分诊，再决定要不要重排

队列构建 40268208875 红了。队列跑的是全量套件（PR 侧 CI 只跑 affected 子集），
所以失败的测试可能在本 PR 没碰过的包里 —— 那不是重排能修的。每次盲目重排都会让排在后面的所有 PR 重建一轮。

分类：failure —— 按下面的日志分诊。

失败的 job（日志抽取，best effort）：

• Test UI (13/16) — 失败步骤: Run the ui tests (shard 13/16)

  × finds no pre-install gate reaching a package 14ms
  FAIL   unit  scripts/__tests__/check-pre-install-import-graph.test.ts > the real tree — every step objectui#6148 measured is still derived > finds no pre-install gate reaching a package
    ↳ 失败原因: AssertionError: a script this repository runs before `pnpm install` needs `node_modules` to load: expected [ Array(1) ] to deeply equal []
  

↳ 失败原因 是判读的关键：超时（Test timed out in … / Hook timed out in …）多半是负载/时序，不是本 PR 的回归；
断言（AssertionError: …）才指向真实的行为改变。两者的 FAIL 行长得一模一样，只有这一行能区分。

⚠️ 断言这一侧有一类例外，判据是断言在测什么，不是它是不是 AssertionError。 断言的对象是产品行为（一个值、一个形状、一次拒收）⇒ 照上面读：真实的行为改变，去查，⛔ 不要重排掉；
断言的对象是这次实验自身的有效性前提（跑完的耗时、负载下的先后、任何只在时间预算内才成立的条件）⇒ 它跟超时是同一类，同样对负载敏感，重排一次是合法的判别手段。
识别是机械的：断言的消息或它比较的值本身点名了一段时长、一个时间戳、一个耗时计数。实测过的一对 —— AssertionError: SecurityPlugin.init() ran: expected false to be true 测的是产品行为（真回归）；
AssertionError: this run took over a second, so second-precision stamps could have differed too: expected 1006 to be less than 1000 测的是实验前提：它守护的那条不变式当时是绿的，同一个 head 原样重排一次即成功。
穿着 AssertionError 外衣的时间测量，仍然是时间测量。（⛔ 这只改「怎么读一次红」，不改「哪些测试可以重排」——后者由别处管。）

跨 PR 相同签名（24h，按失败测试文件聚合）：

• scripts/__tests__/check-pre-install-import-graph.test.ts — 24h 内已弹出 11 个 PR（feat(spec,cli): reports are translatable — reports in the translation bundle #25826、fix(ui-merge): two tests that #25745 broke on main read the merged tree — the Changesets config through changesetConfigPath(), the SDUI html tier by its tags (#25711) #25852、ci(ui-merge): Type Check UI builds the five vitest-setup packages before that step, so an empty ui scope no longer reds it #25856、ci(ui-merge): register Lint UI and Build & E2E UI in REQUIRED_CONTEXTS #25857、fix(ui-merge): the CI page and two of its tests stop reading the two workflows #25745 deleted #25861、test(ui): three ui/ script tests spell their temp bases so the whole-tree scratch-dir sweep reads them (PM dispatch-gates red on main) #25865、ci(ui-merge): register Type Check UI in REQUIRED_CONTEXTS #25878、test(ui-merge): the ui gate-wiring pins read the root ci.yml that runs them; pins on jobs it did not port retire #25879、ci(ui-merge): queue builds decide the four ui lanes from the merged diff against merge_group.base_sha #25881、ci(ui-merge): Test UI runs only the ui tests a pull request or merge group affects — the root ci.yml wires ui/scripts/select-affected-tests.mjs #25882、test(ui-merge): the repo-gates wiring pins retire with their port; "no workflow runs this" pins read the root workflows #25904），分属 2 条投机栈 ⇒ 2 次独立命中。汇总 issue：Queue-flake anchor: scripts/__tests__/check-pre-install-import-graph.test.ts #25863（已刷新）
• ⚠️ 24h 评论账本没读完（超过 5 页仍未读到窗口尽头），所以上面的「不同 PR 数」是下界，不是全量。

历史信号：

• ⚠️ 本 PR 过去 24h 已在队列失败 1 次（不含本次）。 内容未变而反复失败 ⇒ 高度怀疑 flaky 测试或与同组 PR 的语义冲突，重排不解决。
• 过去 24h 队列共有 14 个失败构建（不含本次）。

分诊清单：

1. 失败测试在本 PR 改动的包里 → 真回归，修 PR。
2. 失败测试与本 PR 无关 → 看上面的「跨 PR 相同签名」；已有汇总 issue ⇒ flaky/环境问题实锤，去那张 issue 上谈，修好前重排只会再烧一轮全队列。
3. 两者都不是 → 可能与同组 PR 语义冲突；等前面的 PR 落地或失败出队后再重排一次即可，不要连续重排。

---

Generated by Claude Code · merge-queue-triage workflow (#4859)