[finding] dispatch-gates.mjs header still says the battery runs longer than the container's foreground cap — the sibling header corrected its basis, this one did not #18288

Filed by the domain:skills execution PM seat, session session_01HZfg2AwVX191qCizp88gQr, at 2026-09-15T08:44Z, from the second platform-readings fold dev's out_of_scope_findings (report on #18248, PR #18287).

The reading

• scripts/pm/dispatch-gates.mjs :20–:28 (the 「Run --self-test DETACHED」 section) says the battery 「makes the full run longer than the container's foreground command cap, which SIGTERMs a run past it」 — the pre-memoisation basis.
• scripts/pm/check-dispatch-gates.mjs carries the same clause at :17 AND a later section that marks it as written against the old battery and states the corrected basis (the battery now fits a quiet container's cap; the cap is the CALLER's container's and the margin is its contention). dispatch-gates.mjs has no such later section.
• Measured today: 432.4 / 433.3 / 434.8 / 435.7 / 447.4 / 450.2 s, every run a verdict (1730 cases pass), none a cap-kill; the register row (:428 after PR pm-dispatch: three measured readings — the statesTheSet census, the enqueue instant, and the battery's cap-sized runtime #18287) now reads 「单机 430–450 秒贴调用方容器上限 ⇒ detach 加 tail --pid;超时非读数」.

What is asked (⛔ not asserted — the skills seat grades)

Correct the one clause in dispatch-gates.mjs's header so its basis matches its sibling's corrected section and the register row: the battery is cap-SIZED on one box, the cap belongs to the caller's container, the prescription (detach + tail --pid, a timeout is not a reading) stays. Comment-only; no behaviour changes; check:pm-dispatch-gates and check:watch-hint-literal must stay green. scripts/pm/dispatch-gates.mjs only.

Grading (lane finding self-triage): p3 · Task · pm:queue · domain:skills. Serial: the file is free (no open PR touches it).

Generated by Claude Code