docs(agents): a pin-breaking removal ships with the sibling fix and pin bump

[os-zhuang]

Fixes #10986

What

Encodes the C half of the maintainer's 2026-08-22 A+C ruling (accepted verbatim 「接受所有」) as a working agreement in AGENTS.md, Post-Task Checklist item 4:

A removal that breaks the pinned sibling checkout ships together with the sibling fix and the pin bump — or it does not ship.

Rationale encoded with the directive: the Console Pin Gate job builds objectui at the pinned .objectui-sha against current main, so a removal the pinned sibling still imports turns main red for every PR in the repo the moment it merges. A ruling that authorizes such a removal therefore implicitly authorizes the objectui-side fix and the pin bump as part of the same landing (including the bump's sdui:manifest second half). The item ends with the one-line pre-merge check: does the pinned sibling import what you are removing? git grep it in ../objectui at the pinned SHA.

Placement: new Post-Task Checklist item 4, directly after item 3's retirement paragraph (the prose that points a retirement author at the spec-property-retirement skill) and directly before the spec-export-removal item — the PR-flow section a retirement author actually reads. Former items 4-6 renumbered to 5-7; the only externally referenced item number is item 3 (docs/audits/2026-07-unknown-key-strictness-ledger.md, ADR-0115), which is unchanged.

Line ratchet payment

AGENTS.md is ceilinged at 961/961 with zero headroom. The one added line is paid by deleting the seventeen-phantom-pins incident parenthetical in the Build and Test section (the @ts-expect-error bullet) — incident case law lives in git history, per the ratchet header's own guidance. After the edit: check-skill-line-ratchet: AGENTS.md is 961 lines (ceiling 961; headroom 0) — green, verified at commit e94c5759.

Deferred (declared, not dropped)

The optional one-line pre-merge check in .claude/skills/spec-property-retirement/SKILL.md is deferred: PR #11092 is in flight touching that file with auto-merge armed, and creating a conflict with it is worse than a follow-up. The AGENTS.md item carries the same one-line check, so the rule is not silent in the meantime.

Gates (all at commit e94c5759, exit codes captured before any pipe)

• check:pm-skill-ratchet — AGENTS.md is 961 lines (ceiling 961; headroom 0)
• check:pm-skill-id-lint — 17 file(s) clean (pattern of three-plus-digit issue refs); the directive is self-contained, no issue numbers
• check:doc-authoring — 389 files clean — no bare metadata literals
• check:nul-bytes — OK (scanned 6417 text file(s) ... no raw ASCII control bytes)
• check:docs-audit-scope, check:pm-governed-merges, check:pm-governed-prose, check:required-contexts (both spellings) — all green
• Derivation: node scripts/pm/dispatch-gates.mjs with no hand-fed paths; change set = AGENTS.md only

Governed surface

AGENTS.md is a governed face: this PR stays draft, human merge only — never flipped ready, never queued, never auto-merged.

Generated by Claude Code

---

Generated by Claude Code

[os-zhuang]

PM 验收 — ACCEPT（终态）

复核：席位 session_01MsbKEG4LtERSLaDrbehM3e 全 diff 审毕（+6/−6 单文件）。裁定实现吻合：#10856 A+C 裁定的 C 半逐字落为 Post-Task Checklist 第 4 条，指令 + 单句因由（Console Pin Gate 以 pinned .objectui-sha 对当前 main 构建）+ 一行合前自查（在 ../objectui pinned SHA 上 git grep 被移除面）。指令自含无 issue 编号（id-lint 绿）。重编号 4-6→5-7 安全性经席位独立复核：全仓仅第 3 条被按编号外部引用（audits ledger、filter-operator-vocabulary.test.ts、filter-text-conformance.ts），4-6 条零编号引用。棘轮支付合规：961/961，以删除「十七幻影 pin」事件插叙支付（案例史归 git 历史，棘轮头自述方针），非 re-wrap。退休清单锚点 DEFER 正当（#11092 派发时未合，避免与在飞 PR 冲突；AGENTS.md 条目自带同一行自查，规则不哑）——#11092 落地后锚点行作为后续可选跟进。

治理面终态：AGENTS.md —— 本 PR 保持 draft，仅维护者合并；⛔ 不 ready、不入队、不布防。已指派维护者。合并后剥离 #10986 的 pm:dispatched。

---

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 32587834474 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。

失败的 job（日志抽取，best effort）：

• Console Pin Gate — 失败步骤: Build the Console SPA at the pinned objectui SHA

  ✗ Build failed in 5.22s
  

↳ 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。

跨 PR 相同签名（24h，按失败测试文件聚合）：

• ⚠️ 本次没有可用的聚合签名（日志里没有能解析出测试文件名的 FAIL 行）—— 这不是「没有同签名的其他 PR」，是这一轮没测到。跨 PR 聚合本次不可用，请手工比对其他 PR 的同类评论。
• ⚠️ 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 82 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)