[objectstack-fleet]

Requested by Irene · project thread

Part of #29471

Clause-②: no
Metadata-protocol: none

维护者速读

• 改了什么:删除半态巡检行 H25(pm:awaiting-maintainer 与另一个 pm 状态标签共存),并删掉 H29 里「含 awaiting 的标签对归 H25」那一行跳过,让 H29 报出 H25 原来报的全部标签对。
• 为什么改:scripts/pm grew by about 7,900 lines in a week while the skill text is ratcheted: census the half-state rows and fleet tools by reader, delete what nobody reads, and make the next 30 days net-negative #29471 普查把 H25 列为一档(无读者、30 天无引用);席位裁决选项 C(评论 6589866189):只删 H25,H24 保留。
• 风险与回滚:只动 scripts/pm/check-half-states.mjs;没有哪个状态组合失去报告。回滚即 revert 本 PR。
• 先看:h29PmStateExclusivity 删掉的那行 continue,以及 H29 自测里由「静默」翻成「报出」的七条。
• 两分钟验证:node scripts/pm/check-half-states.mjs --self-test,预期 5202 条全过(主干 5219)。
• 需要你做的:无,席位落地。

What changes

• H25 is deleted: its rationale block, AWAITING_MAINTAINER_CONFLICT_REASON (only H25 read it), h25AwaitingMaintainerExclusivity, its line in the header docblock, its band-registry entry, its call in the sweep loop and the h25row test helper.
• H29 now owns the awaiting pairs. The line if (a === AWAITING_MAINTAINER_LABEL || b === AWAITING_MAINTAINER_LABEL) continue; // H25's is deleted, so H29 reports every pair that contains pm:awaiting-maintainer. This is the one behaviour change to a staying row, ruled by the seat (option C, #29471 comment 6589866189). H29's H3 exclusion (pm:queue + pm:dispatched) is unchanged.
• Kept: AWAITING_MAINTAINER_EXCLUSIVE_LABELS, because PM_EXCLUSIVE_STATE_LABELS spreads it; the queued fixture, which H3, H8, H19, H22 and H29 cases use; H24 and the H8 case that names it, per the ruling.
• Prose pointers that named H25 as a live row now name H29: the AWAITING_MAINTAINER_LABEL reader list, the GOVERNED_PR_DECISION_LABEL note, H55's header, the H66/H67 population notes, the patrol input-set note, the sweep-loop comment and the H66 battery case's name. These are comment and test-name text only, with no output change.
• No gate, row, constant or battery is added. No battery shrank, so no pin moved.

Why

The #29471 census put H25 in tier 1: no named reader and no 30-day citation. The census grouped it with H24 as "the label pairs H29 already covers", but on main H29 skipped every awaiting pair as H25's, and H24 is a label-and-assignee intersection that H29 never reads. The seat narrowed the family to H25 and ruled that H29 takes over H25's pairs, so no state loses its report.

No state loses its report

Measured by importing both trees (origin/main 0a4e74e and this head) and judging every non-empty subset of the six-label ONE-OF vocabulary as an open card:

• 63 label sets judged. H25 fired on 31 of them, and the new H29 fires on all 31 (0 misses).
• On all 63 sets, the answer to "does H3, H25 or H29 fire?" is the same before and after.
• On every set without pm:awaiting-maintainer, H29's sentence is byte-identical before and after.

Verification

• node scripts/pm/check-half-states.mjs --self-test on 608b908: 5202 cases pass (5219 on main, so 17 fewer). The 17 removed are 14 H25 cases (9 single cases and the 5-case per-label loop), H29: …and H25 does fire on it, the H55 adjacency case that called H25, and the H55 reconciliation case that read H25's header prose.
• The seven flipped cases, each from "silent here, H25's" to "fires here": H29: a pair containing the awaiting state is this row's, the 5-case loop H29: awaiting + <state> fires here, and H29: …and names the awaiting pairs too.
• Every battery reads the same n/floor as on main (H66 182/172, H70/H71 31/30 and so on), so nothing was re-pinned.
• Net lines: +35 / −156 (−121), one file.

Local gates on 608b908 (`dispatch-gates --commands`)

• On head 608b908, node scripts/pm/dispatch-gates.mjs --commands derived 47 commands. All 47 ran, each with a 900 s timeout, with exit codes recorded before any pipe. All 47 exited 0.
• The --ran reconciliation printed: 47 derived famil(ies) accounted for — 47 run, 0 NOT-MEASURED.
• node scripts/check-commit-card-trailers.mjs --range origin/main..HEAD: exit 0.
• Prerequisite: pnpm install --frozen-lockfile.
• Families CI runs beyond the derived list are read from this PR's check runs, not re-run locally.

Acceptance notes

• scripts/pm/ensure-pm-labels.sh:212 still names H25 in a comment as a consumer of the label. That file was outside this card's surface, so it was left alone. A one-word follow-up (H25 → H29) belongs to whichever family next touches that file.
• H24 stays. It is the only row that reports an open pm:queue card with an assignee, and H49, H50, H66 and H71 each decline that shape as H24's.

Human Minimum

• 1. Schemas start as Zod; packages/spec holds no runtime logic. (not touched)
• 2. Every name I added uses the casing for its kind. (no names added)
• 3. No formatter ran; pnpm lint passes. (no formatter ran; pnpm lint was not run locally, so the CI lint job decides it)
• 4. One issue, one branch off an up-to-date main, one pull request.
• 5. A change that publishes carries a changeset, and both declaration lines are filled in here and in the changeset. (publishes nothing: scripts/pm/** only; skip-changeset)
• 6. packages/spec touched: its generated artifacts are regenerated and committed. (not touched)
• 7. No hand edits to content/docs/references/, packages/*/CHANGELOG.md or content/docs/releases/.
• 8. Title, body and comments are in English. (except the 维护者速读 block, which the seat contract writes in Chinese)
• The Before Submitting list in CONTRIBUTING.md is done.

🤖 Generated with Claude Code

https://claude.ai/code/session_01J8VVAmM3BRQQcQSA67VE3L

[objectstack-fleet]

domain:skills#1 · huangyiirene · session_01J8VVAnMUWA9rufXzFaEh4J

Contract-review-claim: session_01J8VVAnMUWA9rufXzFaEh4J · head 608b908 · 2026-11-02T22:33Z

The seat review runs at CONTRACT_REVIEW_TIER in an isolated subagent (the seat session is not at that tier).

[objectstack-fleet]

domain:skills#1 · huangyiirene · session_01J8VVAnMUWA9rufXzFaEh4J

Contract review

Served-tier: CONTRACT_REVIEW_TIER
Head-sha: 608b9084462ff4b698413b86888e8675432594fc
Local-runs: probe — on a detached worktree of the head and a second one at the merge-base (origin/main 0a4e74e when the branch was cut; origin/main at 7fc4d23 when fetched; check-half-states.mjs and ensure-pm-labels.sh unchanged on main since the merge-base), Node v22.23.4, no node_modules: node scripts/pm/check-half-states.mjs --self-test on both trees (head 5202 cases exit 0, merge-base 5219 cases exit 0, battery roster and every n/floor byte-identical between the two outputs); a probe script in the scratchpad that imports both trees' check-half-states.mjs and judges all 63 non-empty subsets of PM_EXCLUSIVE_STATE_LABELS as open cards through H3, H25 (base), H29 and H24 of each tree, plus closed, undefined and non-state-label edge cards; node scripts/pm/dispatch-gates.mjs --commands (47 commands, matching the PR body's 47) with every command run from the head worktree in its node form through the head's package.json, 300 s per command: 47 run, 38 exit 0 (among them check:pm-half-states 5202 cases, check:pm-dispatch-gates in 248 s, check:scripts-symbol-anchors, check:self-test-wired both forms, check:pm-post-stamped, check:commit-card-trailers), 9 non-zero, all nine on this container's missing dependency (yaml ×5, typescript ×2, @typescript-eslint/parser ×1 as PREREQUISITE NOT MET exit 3; @changesets/cli ×1 as Cannot find package exit 1 inside check-self-test-workflow-commands), none on the diff, those nine NOT MEASURED locally and read from CI; node scripts/pm/check-governed-merges.mjs --branch HEAD (exit 0, NOT governed, 191 changed lines); node scripts/check-commit-card-trailers.mjs --range origin/main..HEAD (exit 0); git grep for H25, h25, AWAITING_MAINTAINER_CONFLICT_REASON, H24 and h24QueuedWithAssignee over both whole trees; git diff --numstat and a count of removed and added t(/b( lines in the diff. Both worktrees were removed after the runs and the main checkout was not touched.
Scope: scripts/pm/check-half-states.mjs (the one file in the diff: H25's rationale block, conflict-reason map, predicate, band-registry row, sweep-loop call and self-test cases removed; H29's awaiting skip removed; comment pointers and test names re-pointed from H25 to H29) + read, not reviewed: scripts/pm/ensure-pm-labels.sh line 212 (the comment naming H25, outside the surface), scripts/ci/scheduled-full-run.mjs line 536 (H24's one reader outside the file), .claude/skills/pm-dispatch/references/state-machine.md line 8 (H29's named reader), .github/workflows/half-state-patrol.yml (the pull_request trigger on this file and the Live half-state sweep job), .github/workflows/lint.yml line 1265 and package.json line 82 (the check:pm-half-states self-test alias and its step), card #29471 body, comments 6586191438 (census), 6589842500 (claim), 6589866189 (seat ruling), PR #29626 body and its one comment 6590192434 (the seat's review claim on this head)

Reviewer: an isolated contract reviewer at CONTRACT_REVIEW_TIER, read-only on GitHub (REST reads only; nothing written, pushed, commented or labelled), fed the card body and comments, the PR body, the net diff against the merge-base, the head's check-runs, contract-review.md and review-checklist.md. The PR body's measurements were read as input, not as conclusions; every count below is this review's own measurement. No environment variable value is printed anywhere in this record.

① Derived judgments

1. File surface. One commit (608b908, author Claude, PR author objectstack-fleet[bot], head repo = base repo, draft, base main, trailers Co-Authored-By: Claude + Claude-Session:, no model identifier; check-commit-card-trailers exit 0 on the range). git diff --numstat 0a4e74ea 608b9084, the REST file listing and the PR body agree on exactly one path, scripts/pm/check-half-states.mjs +35/−156 (blob f483372c → 1f6d84fe). The claim's file surface (6589842500) was this file plus any skill or reference line naming H24 or H25 as a live row; no skill or reference line names either (measured in 4 below), so the second half is empty. Correct, and on surface.
2. H25 is gone as a row, whole. Removed on the head: the header-docblock entry (15 lines), the rationale block (40 lines), AWAITING_MAINTAINER_CONFLICT_REASON (10 lines, read only by H25 at the merge-base, measured by grep), export function h25AwaitingMaintainerExclusivity (18 lines), H25: 'state' in HALF_STATE_FAMILY_BAND (62 → 61 keys, measured by import), the two sweep-loop lines const doubleState = … / findings.push([issue, 'H25', …]), and the h25row test helper. git grep -E 'H25|h25|AWAITING_MAINTAINER_CONFLICT_REASON' over the head file returns nothing; head.h25AwaitingMaintainerExclusivity is undefined on import. Nothing outside the file imported the predicate or the map at the merge-base. Correct.
3. H29 takes over exactly H25's pairs. The one behaviour line removed is if (a === AWAITING_MAINTAINER_LABEL || b === AWAITING_MAINTAINER_LABEL) continue; // H25's; the H3 exclusion (pmStatePairKey(a, b) === H3_PAIR_KEY) stays. Measured by importing both trees and judging every non-empty subset of the six-label vocabulary (PM_EXCLUSIVE_STATE_LABELS identical on both trees: awaiting, queue, dispatched, blocked, on-hold, needs-user-decision) as an open card: 63 sets; H25 fired on 31 (every set containing the awaiting label and at least one other state); the head's H29 fires on all 31, 0 misses; on every one of the 31 sets the head's H29 sentence names, in backticks, every non-awaiting state present; whether any of H3/H25/H29 fires is unchanged on all 63 sets; H3 is unchanged on all 63; on the 31 sets without the awaiting label the head's H29 return is byte-identical to the merge-base's (null or the same sentence). Edge cards: awaiting alone → null; awaiting + pm:blocking + domain:skills → null (non-state labels still excluded); closed awaiting + queue → null (closed cards stay H22's); undefined → null on both trees. On the triple awaiting + queue + dispatched the head's H29 names awaiting+queue and awaiting+dispatched and still skips queue+dispatched, which H3 reports: one breach, one row, as the H29 header requires. Correct; the PR body's three numbers (63, 31/31, byte-identical off the awaiting label) reproduce.
4. H24 and its readers untouched. h24QueuedWithAssignee's body is byte-identical on both trees (same md5 of the function span); over 126 probe cards (63 subsets × assigned/unassigned) its return is identical on both trees; its sweep-loop call and 'H24' push stay; the five H24: self-test cases and the H8 case …and points at H24 as the state it prevents (head line 30547) are not in the diff. The only lines mentioning H24 that changed are three comment lines that also named H25 (the ## H24 + H25 docblock heading, the GOVERNED_PR_DECISION_LABEL note, the sweep-loop comment). H24's one reader outside the file, scripts/ci/scheduled-full-run.mjs:536, is not in the diff. Correct, per ruling point 2 and 3.
5. No remaining live reference to H25 anywhere in the tree except one comment. git grep -I -E '\bH25\b' over the whole head tree (and over the merge-base) returns exactly one line: scripts/pm/ensure-pm-labels.sh:212, a shell comment that lists the label's "named consumers" as H25 (that exclusivity), H11, H13's PM_STATE_LABELS and H22's PM_RESIDUE_LABELS. Nothing under .claude/**, AGENTS.md, docs/**, .github/** or any other script names H25 as a row; state-machine.md:8 already names H29/H30 as the authority for the six-state ONE-OF, which this PR makes true for the awaiting pairs as well. Judged in ③ and ④: a one-word stale comment outside the surface, no behaviour, correctly recorded in the PR body as a follow-up.
6. The self-test count and its decomposition. Head 5202, merge-base 5219, difference 17. From the diff: 18 t( lines removed, of which 2 are per-label loops over the 5-member AWAITING_MAINTAINER_EXCLUSIVE_LABELS (16 + 10 = 26 cases); 5 t( lines added, of which 1 is a loop (4 + 5 = 9 cases); 26 − 9 = 17. By group: 14 H25 cases (9 singles + the 5-case loop) removed with no replacement; H29: …and H25 does fire on it removed with no replacement; the two H55 adjacency cases (H25 silent, H29 silent) replaced by one (H29 silent): net −1; H55 reconciliation: H25's deferral points at this row removed (it asserted H25's header prose, which the ruling says goes): net −1. The seven flipped cases are the single H29: a pair containing the awaiting state is this row's (was …is H25's, null → 'string'), the 5-case loop H29: awaiting + <state> fires here (was stays H25's row, null → 'string') and H29: …and names the awaiting pairs too (was …does not re-report, false → true); the remaining added line is a rename of the awaitingPlusTwo case with the same assertion. One b( line changed in BATTERY66 (name only, H25's pair → H29's pair, same assertion). Correct, and the PR body's breakdown matches.
7. No battery shrank, no pin moved. The battery roster printed by both self-tests is identical in order and in every n/floor: H66 182/172, H19 37/34 and 26/16, H65 46/42, H67 158/158, H2/H47/H66 213/184, ISSUE_BODY_LIMIT 55/52, H19/H26 11/10, H68 26/24, H9 10/9, lane aliases 25/22, yielded claim 21/19, H69 70/40, H43 89/80, H70/H71 31/30, H72 22/20, H73 28/25. No b( case was removed. Correct.
8. What stays, per ruling point 3. AWAITING_MAINTAINER_EXCLUSIVE_LABELS is still exported with its five members (measured by import) and PM_EXCLUSIVE_STATE_LABELS still spreads it; the queued fixture is still defined at head line 30491 and used by the H3, H8, H19, H22, H24 and H29 cases; SEEN_LABEL_PAGES is unchanged and lists AWAITING_MAINTAINER_LABEL, so every card H25 could have seen is still listed into the same seen loop where H29 runs (head lines 23268 → 23314). The live coverage therefore moves with the predicate, not only the predicate. Correct.
9. Comment and name re-pointing. The AWAITING_MAINTAINER_LABEL reader list, the GOVERNED_PR_DECISION_LABEL note, H55's header (two places), the H66/H67 population notes, the patrol input-set note, the sweep-loop comment and the BATTERY66 case name now say H29 where they said H25 or H25/H29. Each re-pointed statement is true on the head (H29 does own the awaiting pairs now). The H29 header's "both exclusions … in both directions" paragraph was rewritten to the one exclusion that remains; the PM_STATE_CLAIM note about why it is "NOT merged with H25's map" is removed with the map. Correct, and no output string changed by these edits.
10. Gates. check-governed-merges.mjs --branch HEAD: 0 of 1 path hit the register, NOT governed, 191 changed lines, exit 0; the claim and the PR body say the same. Of the 47 derived commands, 38 exit 0 here and reproduce the dev's "all 47 exit 0" for the subset this container can run (the file's own gates among them: check:pm-half-states 5202 cases, check:scripts-symbol-anchors 0 dangling anchors after the deletion, check:self-test-wired with the check:pm-half-states alias still paired to its lint.yml step); the other 9 stop on this container's missing yaml, typescript, @typescript-eslint/parser or @changesets/cli dependency (PREREQUISITE NOT MET exit 3, or Cannot find package exit 1 in check-self-test-workflow-commands), not on the diff, and are read from CI instead: Lint · repo & tooling gates, Lint · CI, release & engine gates, Lint · PM dispatch-gates, Lint · docs & code guards, Lint · scoped · tooling self-tests and Lint & Repo Gates are all success on the head. check:pm-half-states is the alias CI runs for this file (lint.yml line 1265, in Lint · repo & tooling gates, success on the head). The half-state-patrol.yml pull_request trigger fires on this path and runs the live sweep with the head's code (anchor write skipped, body to the step summary); its conclusion is read in the Check-runs line.
11. PR body against the head. First content line Part of #29471, zero verbs beside the card number; the body's only issue number is 29471 (four mentions, none within a closing keyword's reach; a case-insensitive scan for close/fix/resolve forms before #<n> returns nothing); Clause-②: no and Metadata-protocol: none are present as their own lines and right (no published package, no spec, no governed text); PR-body guards check success. The body's numbers (5202/5219, 17, seven flipped, 63/31/31, +35/−156, 47 commands, exit 0 for governed-merges and trailers) all reproduce. The acceptance note on ensure-pm-labels.sh:212 is accurate. No model identifier in body, title or commit. Correct.
12. Nothing new on any surface: no gate, no row, no constant, no label, no workflow, no package. The card asked for deletion "with their self-test cases; no behaviour change in the rows that stay"; the ruling (6589866189 point 1) explicitly accepts the one staying-row change as the merge of two detectors of one state, and 3 above measures that no state loses its report. Correct.

② Semver level

None. No published package, no packages/spec/src/**, no TypeScript; a scripts/pm tool's self-test and predicate. skip-changeset is the right declaration and Check Changeset is success on the head.

③ Boundary flags

• Dev open_questions: none; the R1 report is the PR body (no separate os-dev-report on the card for R1). The census's two open questions (6586191438) were about tier 2 and the sweep as a whole and do not bear on this PR.
• The ruling's narrowing (R1 → H25 only, H24 stays) is implemented as ruled: H24 untouched, the H8 case naming H24 untouched, the H55 header-prose case removed, AWAITING_MAINTAINER_EXCLUSIVE_LABELS and queued kept. No flag.
• scripts/pm/ensure-pm-labels.sh:212: a shell comment still names H25 as the label's exclusivity consumer. It is outside the claim's file surface (a script, not a skill or reference line), it is comment text with no behaviour and no gate reads it, and review-checklist's four conditions for an in-place out-of-surface fix (same defect class, mechanical, unclaimed, same gate family, with the claim's file surface revised in the same round) were not exercised; the dev declared it instead. Judged a follow-up, not a should-fix for this PR: the correct spelling now is "H29 (that exclusivity)", one word, and it belongs to whichever family next touches that file (or a one-line patch after this lands). Recorded here so the tier-2 wave inherits it.
• H29's sentence for an awaiting pair now carries each label's own PM_STATE_CLAIM clause instead of H25's per-pairing "specific lie" clause. The remedy (keep the one true state, drop the rest in a single write) is the same, both labels are still named in backticks, and the ruling accepted the merge; noted, not flagged.
• Independence: PR author objectstack-fleet[bot], commit author Claude, commit Claude-Session: is the dev session (session_01J8VVAmM3BRQQcQSA67VE3L), not the reviewer's; no off-relay author note is owed. The PR thread carries the seat's Contract-review-claim: for this head (6590192434) and no other record.
• Not governed, Tier S path: landing by the ordinary queue once the check-runs are all green.

④ Design judgment

Per hits file, against the card's scenario (a row with no named reader and no 30-day citation costs a self-test and a live pass on every patrol run; the discipline text gives such an instrument deletion as its only in-flight work) and the ruling's constraint (no state loses its report):

• scripts/pm/check-half-states.mjs, H25's removal: cut whole rather than trimmed. The rationale block, the conflict-reason map and the predicate existed only for each other, and the map's prose about "the second half of one sentence" was the reason it was never merged into PM_STATE_CLAIM; with the row gone there is no second half to keep. Leaving the header entry or the band row would leave a registry naming a row the loop never pushes, which the registry-coverage pin would then be asserting for nothing. Holds.
• H29's awaiting skip: the right line to delete, and the only one. At the merge-base H29 already owned the general ONE-OF invariant and the vocabulary constant H25 contributed to (PM_EXCLUSIVE_STATE_LABELS spreads AWAITING_MAINTAINER_EXCLUSIVE_LABELS), so the skip was the one place the two rows met; removing it makes H29 the single owner of 14 of the 15 pairs and H3 of the fifteenth, which is what state-machine.md:8 already said H29 was. Keeping the H3 exclusion is correct because H3 keeps its own measured specimen and sentence. The self-test's seven flipped cases pin the new ownership in the same two-direction form the header demands. Holds.
• The H55 reconciliation case that read H25's header prose through readFileSync(SELF_PATH): it asserted a sentence that no longer exists and could not have been kept without keeping prose about a deleted row; removing it with the row is the ruling's point 3. The remaining H55 reconciliation cases still read the file for the ager's two notes, which are untouched. Holds.
• H24 kept and its one reader (scheduled-full-run.mjs:536) untouched: the census's grouping of H24 with H25 was wrong on the merits (an assignee intersection, not a label pair), the dev measured that before deleting and the seat ruled on it; the PR does what the ruling says and records H24's tier-2 standing in the body rather than in code. Holds.

Must-fix: none. Should-fix: none (the ensure-pm-labels.sh:212 comment is a recorded follow-up, see ③).

Check-runs at record time (2026-11-02T22:49Z): 51 check-runs on 608b908, 30 success, 20 skipped, 0 failure, 1 still in progress on the sixth read (the aggregate Test Core job; both of its shards, 1/8 and 2/8, are success); six reads over thirteen minutes went 14 → 26 → 26 → 29 → 29 → 30 success with no non-success completion at any read. Success: Live half-state sweep (the half-state-patrol.yml pull_request run that executes the head's H29 against the real board, anchor write skipped), Lint & Repo Gates, Lint · repo & tooling gates (the job that runs pnpm check:pm-half-states), Lint · CI, release & engine gates, Lint · PM dispatch-gates, Lint · docs & code guards, Lint · ESLint, Lint · scoped · tooling self-tests, Lint · scoped · slot lookup & corpus, Lint · scoped · query-options erasure, Type Check · workspace / source gates / debt ledger / consumer gates, Test Core (1/8) and (2/8), Test UI, PR-body guards, Check Changeset, Check PR Size, Dogfood Regression Gate, Governed Surface Queue Guard, Plan Test Core shards, Auto Label, Live required-set sweep, Platform checklist gate sweep, Release section-coverage sweep, Snapshot the board to the archive branch, filter. Skipped by path filter or opt-in: Build Core, Build Docs, Build UI Deps, Build & E2E UI, Live E2E UI, Lint UI, Type Check UI, UI Published Artifacts, Dogfood Verify CLI, Temporal Conformance, Schedule gate, 18.x upgrade harness, Packed-tarball smoke, and the matrix placeholders. mergeable_state blocked (draft), auto_merge null, labels size/m, skip-changeset; not governed; the landing pre-check still owes a read of the one in-progress run (the aggregate Test Core job, whose shards are already green on a diff that touches no package source) green before the queue.

Implemented-by: claude/project-thread-22nfer
Reviewed-by: session_01J8VVAnMUWA9rufXzFaEh4J

VERDICT: PASS

---

Generated by Claude Code

[objectstack-fleet]

domain:skills#1 · huangyiirene · session_01J8VVAnMUWA9rufXzFaEh4J

维护者速读

• 改了什么: 从 check-half-states.mjs 删除巡检行 H25,并去掉 H29 对 pm:awaiting-maintainer 组合的跳过，让 H29 报告 H25 原来报告的全部状态组合。净 −121 行，一个文件。
• 为什么改: scripts/pm grew by about 7,900 lines in a week while the skill text is ratcheted: census the half-state rows and fleet tools by reader, delete what nobody reads, and make the next 30 days net-negative #29471 的普查：两个检测器盯同一类状态冲突;合成一个，不丢任何一种状态的报告。H24 经实测没有其他行覆盖，按席位裁定(6589866189)保留。
• 风险与回滚: 63 种状态组合逐一比对:H25 报的 31 种新 H29 全报，其余组合是否报警不变;不含 awaiting 的组合 H29 文案逐字不变。可整 PR revert。
• 先看: H29 去掉的那一行 continue,以及翻成「会报」的 7 个用例。
• 两分钟验证: node scripts/pm/check-half-states.mjs --self-test,5202 例全过。
• 需要你做的: 无需审批(非受治理面，本席落地)。
• 席位意见: ACCEPT。达档复核 PASS(6590474850)。

席位核对 — ACCEPT · domain:skills#1 · head 608b9084462ff4b698413b86888e8675432594fc

1. 形态: draft → main,Part of #29471(卡在 scripts/pm 30 天净变化转负前不关);skip-changeset;作者 objectstack-fleet[bot];一个文件 +35/−156,与认领一致。
2. 前提实测: 认领时的前提(H29 覆盖 H24/H25)经 dev 实测只对 H25 成立;已按裁定收窄。复核方自写探针独立复现 63 组合比对;H24 函数体逐字未变。
3. 门禁 check-runs(钉 head 608b9084): 51 条全部完成,success 31 / skipped 20 / failure 0,含 Live half-state sweep(用新 H29 实扫看板)。check-governed-merges:非受治理。
4. 接受的偏差: scripts/pm/ensure-pm-labels.sh:212 一行注释仍写 H25,在文件面外，正文已声明;由下一个碰该文件的家族顺手改。
5. 范围外去向: H24 的「无读者、无引用」记录在卡上，留给第二波。R2 起改同一文件，待本 PR 合入后串行开始。
6. 落地条款: 非受治理。本席 pr_ready + automerge_enable。

---

Generated by Claude Code