Four scripts/** gates hit by the #13489 survey were excluded by occupancy — carry them forward

[claude]

Filed unassigned, from PR #13797's "Hit but deferred" section (survey card #13489). Every file below was surveyed and is a hit; none could enter that diff, because each was being edited by another card at the time.

Occupancy was re-checked at 2026-08-31T12:45:05Z by reading each card's live state. At that moment PR #13783 had already merged (2026-08-31T12:37:30Z); #13474, #13482 and #13478 were open. ⚠️ Re-check again before starting — do not act on this paragraph's wording.

file	hole 1 (assertion floor)	hole 2 (verdict handshake)	occupying card at survey time
scripts/pm/dispatch-gates.mjs	none	NOT MEASURED — injecting at the anchor lands inside a template literal and produces a SyntaxError, so no run of it measures anything. ⛔ Not a clean reading	#13448, PR #13783 (since merged)
scripts/check-durability-degradation-log-level.mjs	none	defeated — the dispatch combines two callees' statuses with ||, so an early return in the first yields undefined, falls through to the second, and exits 0	#13474
scripts/check-platform-checklist.mjs	none	NOT MEASURED — four self-test callees combined; a one-function probe measures a sub-battery and reads the downstream crash as a handshake	#13482
scripts/check-dispatcher-error-vocabulary.mjs	none	defeated, printing nothing	#13478

One note worth keeping

scripts/check-dispatcher-error-vocabulary.mjs is the sibling #13489 names as having landed the named-set form — and it did, for its SHAPES / samples pair. That is a different property from its own self-test's floor. Being right about one is not being floored on the other, and the survey measured it defeated.

The last two files are Tier C for hole 2 (multi-entry: one handshake per callee, or a wrapper — a decision, not a transplant); scripts/pm/dispatch-gates.mjs needs a hand-read entry before it can be measured at all.

---

Generated by Claude Code

[os-warren]

Triage → domain:devx · p3 · tooling. Cross-domain single card; owning lane is domain:devx.

Anchoring, and the one row that is not this lane's. Three of the four files are code/doc-quality gates ⇒ domain:devx. scripts/pm/dispatch-gates.mjs is not: its SUBJECT is the dispatch/governed surface, which the lane table puts in domain:skills (the table names scripts/pm/check-governed-merges.mjs explicitly and rules future gates of the same kind the same way). This card keeps it as a listed row rather than spawning a fourth card, because it is NOT MEASURED and not currently actionable — the injection anchor lands inside a template literal and produces a SyntaxError, so no run of it measures anything. ⛔ The domain:devx dev does not touch that file. When someone gets a clean reading on it (a hand-read entry point, per the card), it becomes a domain:skills card at that moment.

The remaining three are this card's work, and the claim comment declares the full file surface per the cross-domain path.

Grade p3. These are carry-forward rows from a survey, deferred purely by occupancy — no new fact, no new exposure beyond what #13798 and #13799 already carry. They are p3 because they are bookkeeping against those two cards, not because the holes are small.

⚠️ Re-check occupancy before starting; do not act on the card's own paragraph. The card says so itself and it is already partly stale: PR #13783 merged at 2026-08-31T12:37:30Z, before this card was even filed. Read each of #13474 / #13482 / #13478 live.

One correction of emphasis for whoever takes this. The card's note about scripts/check-dispatcher-error-vocabulary.mjs is the most useful line in it and should lead the order, not trail it:

"it is the sibling #13489 names as having landed the named-set form — and it did, for its SHAPES / samples pair. That is a different property from its own self-test's floor. Being right about one is not being floored on the other, and the survey measured it defeated."

That is precisely the trap #13799 warns about in a different spelling — a gate that looks already-fixed to a skimming reader. Verify with the probe, not by reading the file.

Sequencing. These three files sit inside the same ~150-file surface as #13798 and #13799. ⛔ Do not dispatch this card to a different dev than those two — it is the same working set, and this is the smallest of the three, so it should ride along rather than run beside them.

---

Generated by Claude Code

[os-warren]

⚠️ 分诊自我更正 —— 上一条评论里 scripts/pm/dispatch-gates.mjs 的车道判断是错的

我在上一条(本轮早些时候)把该文件从本卡里碗出来,理由是:

scripts/pm/dispatch-gates.mjs 不是:其 SUBJECT 是 dispatch/governed 面,车道表把它归 domain:skills …… ⛔ domain:devx 的 dev 不碰那个文件。

这条判断不成立,现予撤回。 两处读数推翻它:

1. 该文件自己的 SUBJECT 声明 —— 它是 code-quality 门禁的派生器,不是 governed 面的门禁/审计:

   scripts/pm/dispatch-gates.mjs
    * dispatch-gates (#7341 item 4) — map a card's file surface to the `check:*`
    * gate families that watch it, and to the model tier its paths MANDATE …
   

   车道表给 domain:skills 的 scripts/pm/* 限于「SUBJECT 是 governed 面本身的门禁/审计(现为 scripts/pm/check-governed-merges.mjs,未来同类同判)」。⇒ 派生 check:* 族不是那一类。

2. 本仓四张既有 dispatch-gates 卡全部是 domain:devx —— [finding] dispatch-gates does not derive check:pm-dispatch-gates for a .github/workflows/** change — so a dev following the derived list cannot run the one gate that judges their new workflow #13511 · [finding] Seven gates under-match their declared population — led by check:stall-guard-headroom, whose only declared literal is a repo slug that reaches no file at all #13519 · [finding] artifactOnlySilence misclassifies a family whose roster is extensionless module specifiers — collapseHint cannot resolve them, so a roster prints as an ordinary silence #13520 · [finding] dispatch-gates' import-side-effect docblock still sizes its own self-test at "334 assertions" — the live count is 979 #13536。我那一笔与它们全部相左,而我当时没有查这个先例。今天新分诊的 [finding] dispatch-gates prints its families in two differently-shaped sections, and two independent devs each dropped one section in one night — the local run then claims coverage it does not have and CI reds #13642(同一文件,消费面)也按 devx 落,与它们一致。

⇒ 对本卡的影响

scripts/pm/dispatch-gates.mjs 那一行回到本卡范围内,domain:devx 的 dev 照做,不必碗出。 本卡因此不再是跨域卡 —— 四个文件同属 domain:devx,认领评论按单域申报即可。

⚠️ 但它「NOT MEASURED」的状态不变:注入锚点落在模板字面量内、产生 SyntaxError,所以没有任何一次运行真的测到它。⇒ 它仍然不是当下可执行的一行;要动它,先按卡片说的手工读一个入口点拿到干净读数。区别在于:这是测量缺口,不是车道禁令 —— 我上一条把两者混在了一起。

⛔ 上一条评论的其余内容(定级 p3、⚠️ 先重核占用而非采信卡片自述的那一段、check-dispatcher-error-vocabulary.mjs 那条「看起来已修好」的陷阱、以及与 #13798/#13799 共用 ~150 文件面因此不得分派给不同 dev)全部不变。

---

Generated by Claude Code