Repository navigation
queue: no reconciliation between open issues and queue coverage — non-creator filings never get queue items #7016
Description
Problem
Nothing reconciles the open-issue set against the fleet queue. Issues enter the queue only when a creator-class lane files them (file + enqueue in the same pass). Issues filed any other way — by work lanes filing blockers mid-task, by audit lanes, by hand, or predating the queue (epics) — silently never get a queue item, so no lane ever dispositions them.
Evidence — reconciliation run 2026-11-07 (Drew's question: "did you import all open issues without a queue item?")
Method: full open-issue enumeration via gh REST (PRs excluded) for the four fleet repos; full fleet-queue scan of the four pilot queues (8,493 tasks) matched on dedupe keys repo-health:v1:(fix|triage):<repo>:issue:<n> incl. dispatch-suffixed variants and legacy repo-name aliases.
| repo | open | live task | done-task only | NO task |
|---|---|---|---|---|
| developing-today/code | 1,347 | 656 | 676 | 15 |
| developing-today-forks/opencode | 95 | 44 | 45 | 6 |
| developing-today-forks/openchamber | 95 | 59 | 36 | 0 |
| dezren39/darwin-nix | 30 | 19 | 9 | 2 |
The 23 uncovered issues were imported the same day as triage tasks 11424–11446 (all verified in-store; several already claimed within minutes). They include today's own filings (#6884, #6935, #6942, #6944, #6985, #6992, #6994), home-cluster epics (#1083, #1104–#1106), fork pr-shepherd items (opencode #1715, #1757, #1947), and darwin-nix #11/#1637 — proof the gap is structural, not one lane's miss: any filing path that isn't a creator run leaks.
The 766 "done-task only" issues (worked, issue still open) are a separate, larger lag class — triage/fix tasks complete without the issue closing — flagged here for visibility, not as part of this fix.
Fix
- Standing reconciliation: run this exact diff (open issues vs pilot-queue dedupe coverage) on a cadence (the hourly orchestration pass or its own job) and auto-import uncovered issues as triage packets via the sanctioned enqueue path (dedupe-key idempotent by construction).
- File-and-enqueue rule: any lane that files an issue outside a creator run enqueues its triage packet in the same pass, or records why not in the filing comment.
Acceptance
- A scheduled reconciliation exists and its first run reports 0 uncovered open issues across the four repos.
- The file-and-enqueue rule is written into the lane/creator runbooks it applies to.
Activity
dezren39 commentedon Nov 8, 2026
🤖 Agent note · handed off
developing-today commentedon Nov 8, 2026
🤖 Agent note · handed off
dezren39 commentedon Nov 8, 2026
🤖 Agent note · handed off
dezren39 commentedon Nov 8, 2026
Triage doctrine — refinement (Drew, 2026-11-07 20:32 CST)
Refining the triage doctrine recorded earlier tonight, as the standing rule:
The rule as recorded:
169 remaining items
dezren39 commentedon Nov 9, 2026
agent-note:v1
state: handed-off
queue: repo-health-code-pilot
task: 11447 (triage, worker repo-worker-code, lane code-m), queue attempts 8/12
by: repo-worker-code lane code-m
at: 2026-11-09T05:18:00Z
Handed off to the opencode execution lane (assigned_lane=opencode via fleet queue lane, lane row 3576). Fresh check this shift: issue #7016 OPEN, no hard-stop labels; PR #7697 OPEN at head b135676, authored by dezren39, carrying NO labels - per the fleet rule (Drew, 2026-11-07) authorship alone is not a hold, so PR #7697 is ordinary work, not a no-touch artifact. It is, however, the active implementation for issue #7016 (standing issue-queue coverage reconciliation), and prior progress records its test evidence (cargo test -p fleet-queue --all-features 100/100 lib + 2/2 integration; test_check_coverage.py 6/6). For the claimant session: shepherd PR #7697 to merge under the standing criteria (checks + Codex gate), then record the triage verdict on issue #7016 and complete task 11447; do not build a parallel implementation. This lane does not work task 11447 directly. Parked 1200s to release the claim for claimant binding. Labels synced for claim then park.
dezren39 commentedon Nov 9, 2026
🤖 Agent note · handed off
opencode(lane row 3576)fix(queue): disposition Codex findings on coverage reconciliation (#7016
feat(queue): standing issue<->queue coverage reconciliation (#7016) (#…
dezren39 commentedon Nov 9, 2026
agent-note:v1 task 11447 completed by lane code-l as already_done: issue #7016 is CLOSED (completed, 2026-11-09T05:43Z) and PR #7697 feat(queue): standing issue-queue coverage reconciliation was MERGED at the same time (merge sha e74e657). The triage premise is resolved.
dezren39 commentedon Nov 9, 2026
🤖 Agent note · done (already_done)
opencode(lane row 5526), but fresh revalidation showed the work already landed, so this lane completed it directly per the runbook's already-done outcome.