18 of 71 open `domain:spec` `pm:queue` cards say in their own body that their disposition is a maintainer decision — they read as dispatchable and are not

[os-bill]

⛔ Filed unlabelled and ungraded by the domain:spec execution seat, session session_01MkQhmuuJAVDjmeWNixwDDH, 2026-09-12T10:02Z. domain:*, type and priority are triage's. ⛔ Not claimed, ⛔ not dispatched, and ⛔ this seat answers none of the questions the cards below raise.

The pattern, and why it costs something

A card sitting in pm:queue reads as dispatchable. A card whose disposition is a maintainer-floor decision is not. When both look identical on the board, every dispatcher that reaches one of the second kind does one of two things:

• skips it silently — leaving no record, so the next dispatcher rediscovers it from scratch; or
• dispatches it — and the round hits the fence on its first measurement and stops, which is the expensive outcome.

Both happened in this lane today. Three cards were handed back individually with pm:retriage (#17598, #16066, #17534) after being read one at a time during card selection. ⭐ That is the reactive form of this card: it fixes one card per encounter and never reaches the rest.

Measured, 2026-09-12T10:0xZ

Over every open domain:spec + pm:queue card — 71 of them, enumerated by label rather than sampled:

18 carry, in their own body, a sentence naming a maintainer-floor or decision disposition, and none of the 18 carries pm:retriage.

14512  15124  15932  16075  16160  16282  16348  16524  16870
16906  17108  17109  17158  17296  17319  17321  17518  17594

The probe is a phrase match over the card bodies: maintainer ruling · is a contract decision · not a door fix · ruling-grade · maintainer decision · narrows a published accept set · 人工地板 · is a decision · not_planned, case-insensitive, pm:retriage carriers excluded.

⛔ 18 is an UPPER BOUND, and here is a verified false positive

⭐ A string count is not a reading of meaning, and this probe demonstrates it. #17594 is in the list because it contains ruling-grade — but the sentence it appears in is the card excluding that question from itself:

⛔ No claim the conversions should delete the node. That is a ruling-grade question … This card is about the todo, ⛔ not the conversion's behaviour.

⇒ #17594 is mechanical and dispatchable; this seat has already read it, confirmed that, and is holding it only for file contention on packages/spec/src/migrations/registry.ts. At least one of the 18 is a false positive because the phrase appears in a sentence that fences the floor OUT. Others may be. ⇒ the list is a worklist to read, ⛔ not a verdict on 18 cards, and ⛔ nothing here should be re-graded from the probe alone.

Two of the 18 are already known to this seat by reading rather than by probe: #16282 (the real repair is a published accept-set change) and #15932 (its own text: "retiring a packages/spec surface is a maintainer ruling plus the spec-property-retirement playbook").

The ask

⛔ Not a re-grading request on any specific card, and ⛔ not an opinion on any disposition. The ask is that triage decide whether this population wants a sweep — read each of the 18, and for the ones that really are floor-shaped, move them out of the dispatch queue (decision box with the four-facet block and 維護者速讀, or not_planned, or a split where only part is floor-shaped, as #17598 needs).

⚠️ If the answer is "no sweep, keep handing them back one at a time", that is a legitimate answer and this card closes not_planned — the value was in measuring the population, which had never been counted.

What this card does NOT claim

• ⛔ No claim that any of the 18 is wrongly graded — the probe cannot support that and the false positive proves it.
• ⛔ No claim about cards outside domain:spec + pm:queue; the enumeration is bounded to that label pair and to open cards, read at 2026-09-12T10:0xZ.
• ⛔ No claim that the decision box is the right destination for any particular card; split and not_planned are live options per card.
• ⛔ No severity, no priority, no domain:*.

---

Generated by Claude Code