You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
impl(aggregator/collector): Project pool resolves a minority of board/roadmap refs (172 fixture fallbacks) - raise coverage and publish the ratio #1141
Evidence (live review 2026-10-26 ~22:40 CDT, laptop loopback + gh, cross-checked): Footer tonight: "items: live (project pool - fleet feed - 3m old) + fixtures for 172 ref(s) outside the pool". The board has 107 cards and the pool 139 items, yet most GitHub issue/PR states layered onto cards still resolve from FIXTURES - invented states presented with the same badges as live ones. A reader cannot tell which badges are real.
Proposal
(1) Collector/project: widen pool resolution (the pool should carry the refs the board/roadmap actually link - backfill via the orchestrator's label-gated auto-add plus a one-time sweep of board/roadmap refs). (2) Aggregator: fixture-resolved badges render visually distinct (e.g. hollow + "unverified") until coverage is near-complete. (3) /api/health publishes resolved-live vs fixture counts as a tracked metric.
Acceptance
Fixture-resolved items are visually distinguishable on board/roadmap/item pages.
pm-stack/aggregator/src/sources/live.tsitems() (~lines 695–737): refs resolve pool-first (fleet-feed Project pool, design+impl: org-level GitHub Project as the conceptual-project layer — orchestrator-synced, dashboard-read #910 stage 5), then the projector — whose own code comment says it is "dead by config" — then fix.items(remaining) fixtures. All three are merged into one Map<RefKey, GhItem> via out.set(k, v) with no per-item resolution source. The only trace is a facet-level footer string ("+ fixtures for N ref(s) outside the pool"), so a fixture-resolved badge is indistinguishable from a live one on the board/roadmap/item pages.
pm-stack/aggregator/src/health.ts: the health payload carries no resolved-live vs fixture counts and no coverage ratio for the items facet.
Route: enqueued as queue task 3032 (small_fix, dedupe repo-health:v1:fix:developing-today/code:issue:1141): track the per-item resolution source through items() (pool | projector | fixture), render fixture-resolved items distinctly (hollow + "unverified") on board/roadmap/item pages, publish resolved-live vs fixture counts + ratio in /api/health, with tests for all three. The issue stays open; this triage task completes as triaged.
The re-run (40985792170) remained QUEUED for the full 5-minute poll this shift (~40 min queued total) — runner backlog, not a code failure. Task re-parked 20 min.
Queue task: 3032 (Lane D, repo-worker-code), second resume. Run 40985792170 (re-run of the required "basic: proxy port fallbacks" check on PR feat(aggregator): per-item resolution source — unverified fixture badges + health coverage ratio (#1141) #2425 head 02ee622) is STILL queued ~80 minutes after it was created; an earlier run of the same check on the same head concluded SUCCESS, Codex is clean, and the PR is MERGEABLE — this is runner backlog, not a code failure.
Activity
developing-today commentedon Oct 27, 2026
CODE-278
dezren39 commentedon Oct 29, 2026
🤖 Agent note · done
bd88019448ad397cdbcadaf8e0d29171fb1d661dEvidence (fresh GitHub reads at HEAD):
pm-stack/aggregator/src/sources/live.tsitems()(~lines 695–737): refs resolve pool-first (fleet-feed Project pool, design+impl: org-level GitHub Project as the conceptual-project layer — orchestrator-synced, dashboard-read #910 stage 5), then the projector — whose own code comment says it is "dead by config" — thenfix.items(remaining)fixtures. All three are merged into oneMap<RefKey, GhItem>viaout.set(k, v)with no per-item resolution source. The only trace is a facet-level footer string ("+ fixtures for N ref(s) outside the pool"), so a fixture-resolved badge is indistinguishable from a live one on the board/roadmap/item pages.pm-stack/aggregator/src/health.ts: the health payload carries no resolved-live vs fixture counts and no coverage ratio for the items facet.Route: enqueued as queue task 3032 (
small_fix, deduperepo-health:v1:fix:developing-today/code:issue:1141): track the per-item resolution source throughitems()(pool | projector | fixture), render fixture-resolved items distinctly (hollow + "unverified") on board/roadmap/item pages, publish resolved-live vs fixture counts + ratio in/api/health, with tests for all three. The issue stays open; this triage task completes astriaged.40 remaining items
dezren39 commentedon Oct 30, 2026
🤖 Agent note · parked
basic: proxy port fallbacksre-run 40985792170 stayed QUEUED through the full 5-minute poll window (an earlier run of the same check on this head concluded SUCCESS). No Codex review findings exist on the PR. Parked 20 min per the CI rule.dezren39 commentedon Oct 30, 2026
Agent note — picked up (resume)
dezren39 commentedon Oct 30, 2026
Agent note — parked (CI runner backlog)
dezren39 commentedon Oct 30, 2026
Agent note — parked again (check still queued)
dezren39 commentedon Oct 30, 2026
Agent note — parked on merge queue (lane C, 2026-10-29)
02ee622e, all checks SUCCESS, Codex clean (0 findings), MERGEABLE — and enqueued in the merge queue (position 29, AWAITING_CHECKS).