Repository navigation
pm-stack: Project #4 coverage report — pool vs backlog, gate-label hygiene, and a ruling on the next gate #1220
Description
Part of epic #1184 (PM stack coherence review, 2026-10-26).
Nobody owns Project #4 coverage versus backlog reality. Evidence: the pool holds ~90 items while code alone has 300+ open issues and nix 300+; entry is gated on now/next/epic labels that are barely applied (next = 0 open in every repo sampled, now = 21 total across 4 repos). Unset Status/Area values are common enough that aggregator code special-cases '(unset)'.
Acceptance:
- A first coverage report is published (issue comment or /projects panel): open issues per repo, gate-label counts, unset-field count for the pool, with week-over-week delta if recurring.
- The next gate is either in real use or removed from the orchestrator contract, with the ruling recorded.
- Unset-field count for the pool is known and trending.
Acceptance criteria
- A first coverage report is published (issue comment or /projects panel): open issues per repo, gate-label counts, and unset-field count for the pool, with week-over-week delta if recurring.
- The
nextgate is either in real use or removed from the orchestrator contract, with the ruling recorded. - The unset-field count for the pool is known and trending.
Activity
developing-today commentedon Oct 27, 2026
CODE-354
dezren39 commentedon Oct 28, 2026
Intake review
Related: epic #1184; siblings #1218 (auditability/export), #1224 (staleness lane), #1226 (Needs-Drew queue). The orchestration pass already produces much of the raw material (its run log records Project reconciliation each cycle).
Notes: the report is cheap once someone owns it — this issue is that ownership. The
next-gate ruling (0 open items carriednextat review time) is the decision half. No merged PR references this issue (checked this run).Risks: a one-off report rots; acceptance allows a comment for the first edition but the delta clause pushes toward recurrence.
Suggested approach: generate the first report from the orchestrator's existing census, post it here, and record the
nextruling in the orchestrator contract.🤖 Agent note · filed
35 remaining items
dezren39 commentedon Nov 1, 2026
🤖 Agent note · parked
8da7c9c477still has no Codex review of the current head; GitHub reports mergeability unknown. Parked 20 minutes. Resume by rechecking review, CI, mergeability and the issue acceptance criteria.Merge pull request #3309 from developing-today/queue/4931-1220-covera…
dezren39 commentedon Nov 1, 2026
🤖 Agent note · done — coverage report published, PR merged
/projectsshows "Coverage — pool vs backlog (pm-stack: Project #4 coverage report — pool vs backlog, gate-label hygiene, and a ruling on the next gate #1220)" (per repo: open issues, now/next gate counts, pool items, unset Status, unset Area, Δ vs baseline) and/api/projects.jsoncarries the same report; the collector censuses the four repos via the Search API and the weekly snapshot (with week-over-week delta) is stored beside the other aggregator stores.First report numbers (triage 2496, Search API, 2026-11-01): open issues — developing-today/code 1122, opencode 66, openchamber 54, dezren39/nix (see triage note above); gate labels — code now 22 / next 22; opencode, openchamber, nix 0 / 0. The panel now produces these continuously, including the pool unset-field counts.
Next-gate ruling (recorded, matches the numbers):
now/nextthere.nextset in those repos is not a defect and must not block; a repo opts in by actually labelling, and the panel shows it from that week on. Removing the labels there is not warranted.This shift's closing work: Codex's round on the merged head found one more real P2 — rotation promoted the refreshed
currentto baseline, so the week-over-week delta measured since the last refresh (+10 in Codex's scenario) instead of the week (+70). Fixed in 585d16d (frozenweeklyreport promoted at rotation; regression tests pin the exact scenario; coverage suite 18/18). Main was merged twice more (feed.ts conflict resolved keeping both changes); Codex clean on the final head 3338397.Delta mechanism: the stored snapshot holds
{baseline, current, weekly, anchorAt}; the first report has no prior (baseline null), and each week's frozen report becomes the next baseline — week-over-week deltas exist from the first rotation on.Closing this issue: all three acceptance criteria are met (report published as panel + this summary, delta mechanism stored, ruling recorded and matching the numbers).