Background — the gap, measured on 2026-08-09
The release board (target:v17) has exactly one producer: the objectstack triage seat, whose step-0 sweep is deliberately repo-scoped to objectstack (the SKILL itself warns that an unscoped sweep would drown in sibling-repo queue cards). The board design says "objectui enters the board via the console bundle" — but that clause has consumers and no standing producer on the objectui side:
- objectui's
target:v17 label exists and was used exactly 7 times (the 2026-08-06/07 release-board audit); all 7 are closed/fixed.
- Since then objectui's local backlog grew to 117 open / 48 pm:queue with zero open board items — no one ever asks the binary question ("can the current RC ship without fixing this?") over that backlog, while the console ships from it.
Maintainer ruling (2026-08-09, quoted verbatim)
The PM proposed three actions (add a producer / one-time clearing pass / fix the release-day query scope); the maintainer ruled 「1 补上生产者」 and then 「三个动作都创建issue」 — file all three as issues. This card is action 1; siblings are filed separately and cross-linked.
Deliverable
Amend .claude/skills/pm-dispatch/SKILL.md, 发版板 section, by PR (taxonomy evolves by PR):
- Producer scope becomes "exactly one producer per backlog" (same single-producer discipline, scoped correctly): the objectstack triage seat keeps producing for the main backlog; the objectui whole-repo seat produces
target:v17 for objectui's own local backlog, applying the same binary criterion + four blocking classes during its local backlog sweep. Other execution seats still never add/remove the label; mislabels go through escalation.
- Record the stock-clearing posture: objectstack was cleared once on 2026-08-06 (309 → 46); objectui's stock gets one equivalent pass (companion issue), after which both sides are incremental-only, never full-rescan.
- The sibling-repo line points at the new producer, and notes the release-day inventory is two queries (see the companion query-scope issue).
Guardrail note: this is a .claude/ internal-tooling PR — per the standing exception it may be implemented by a PM seat only with the maintainer authorization quoted in the PR body, and review requires another seat or a maintainer walkthrough; no self-review-self-merge.
No domain:* label applied — routing is the triage seat's single-producer territory.
Background — the gap, measured on 2026-08-09
The release board (
target:v17) has exactly one producer: the objectstack triage seat, whose step-0 sweep is deliberately repo-scoped to objectstack (the SKILL itself warns that an unscoped sweep would drown in sibling-repo queue cards). The board design says "objectui enters the board via the console bundle" — but that clause has consumers and no standing producer on the objectui side:target:v17label exists and was used exactly 7 times (the 2026-08-06/07 release-board audit); all 7 are closed/fixed.Maintainer ruling (2026-08-09, quoted verbatim)
The PM proposed three actions (add a producer / one-time clearing pass / fix the release-day query scope); the maintainer ruled 「1 补上生产者」 and then 「三个动作都创建issue」 — file all three as issues. This card is action 1; siblings are filed separately and cross-linked.
Deliverable
Amend
.claude/skills/pm-dispatch/SKILL.md, 发版板 section, by PR (taxonomy evolves by PR):target:v17for objectui's own local backlog, applying the same binary criterion + four blocking classes during its local backlog sweep. Other execution seats still never add/remove the label; mislabels go through escalation.Guardrail note: this is a
.claude/internal-tooling PR — per the standing exception it may be implemented by a PM seat only with the maintainer authorization quoted in the PR body, and review requires another seat or a maintainer walkthrough; no self-review-self-merge.No
domain:*label applied — routing is the triage seat's single-producer territory.