os-dev.md's Part-of bullet lacks the bookkeeping-sentence / quote-backtick shapes that PR #13505 sharpened in review-checklist.md — deferred by the frame-region fence

[zhuangjianguo]

Filed by the skills lane seat (session session_01EXxTW8mvPBhoHxmyPZ63de, PM loop round 3), routing a fence-respecting deferral declared in the #12668 dev report. Lane self-triaged: pm:queue · domain:skills.

The gap

PR #13505 (the #12668+#12779 landing) sharpened the closing-keyword rule at the Part-of switch point in .claude/skills/pm-dispatch/references/review-checklist.md: the reference parser reads the who-will-retire-it bookkeeping sentence, negations, and quote/backtick wrappings all the same way — nowhere in a PR body may a closing keyword sit adjacent to a card number — with the safe spelling given (bare card number with zero closing verb; the closing plan goes into a comment).

.claude/agents/os-dev.md carries its own Part-of bullet, which already states the prohibition, the negation fact, and three safe spellings — but NOT the bookkeeping-sentence and quote/backtick shapes. That bullet sat outside the frame-copy region the #12668 dispatch was fenced to, so the dev correctly deferred it instead of widening its surface.

Constraints at claim time (re-derive, never recall)

• os-dev.md is governed (.claude/**) and line-ratcheted — 466/466 today, with a RULED raise to 471 pending on [finding] A timeout-killed command reports near-zero CPU to the shell's time, understating cost 14x-82x — and it reads exactly like "this step blocks rather than computes" #12980's landing (决裁批 [WIP] Create a new release version #9). Whoever takes this card lands AFTER [finding] A timeout-killed command reports near-zero CPU to the shell's time, understating cost 14x-82x — and it reads exactly like "this step blocks rather than computes" #12980 and re-reads the ceiling then; the edit is one parenthetical extension of the existing bullet, in-place word compression only, ⛔ no re-wrap funding (2026-08-17/2026-08-29 ruling pair).
• Natural fold candidate: whichever card next opens os-dev.md's non-frame prose after [finding] A timeout-killed command reports near-zero CPU to the shell's time, understating cost 14x-82x — and it reads exactly like "this step blocks rather than computes" #12980 ([finding] The re-wrap funding ban lives only in the PM's SKILL.md — two devs paying line ratchets hit it independently in one hour because the dev contract never shows it #13116 · Make check:generated's "not run here" source-audit list a hard obligation in the os-dev contract — disposition 2 of #13207, routed to the skills lane #13300 · [finding] .claude/agents/os-dev.md rule 1 lists the verify lock's four guarantees and no non-guarantee, one line above a wall-clock absolute — every dev agent reads exclusivity the brief never promised #12800 · [finding] A PR-body PATCH DELETES the whole attribution footer block, not "normalises it to bare" — AGENTS.md:481 and os-dev.md:463 both state the measured-false behaviour, and three closed cards banked nothing #12886 · [finding] The ablation restore leg is provable only by a WHOLE-TREE status — an agent that reasons about it without running the preflight has no rule telling it so #13293 queue) — fold-or-serial answered at dispatch time.

Refs

#12779 (the evidence card: PR #12777 measured red-then-green; PR #13258 prevention case) · #12668 / PR #13505 (where the sharpened wording lives) · #12980 (the pending ruled raise on this file).