Repository navigation
pm stack → Supabase phase 4 (#3142): backlog state derived from DB + GitHub, markdown becomes a view #5007
Description
Phase 4 of epic #3142: backlog state derived from DB + GitHub (markdown becomes a view)
Part of epic #3142 (Long-term: migrate the pm stack into Supabase). This is the implementation issue for phase 4; phases 1, 3 and 5 are filed separately, and phase 2 already has implementation issues (#1744, with #941 in its area).
Scope
- Backlog state (the
backlog-*.mdfiles the orchestrator reads in the goal workspace) is derived from Supabase + GitHub state instead of being hand-maintained; the markdown files become a generated view. - Queue/backlog consumers keep reading the same markdown shape during the transition; the flip happens only after the derived view matches the hand-maintained files.
Dependencies / blocked-by
- Blocked by the merged migration plan preconditions in
pm-stack/SHARED-SCHEMA-MIGRATION-PLAN.md(PR pm-stack: per-service migration plan for the shared schema (#1744 slice 2) #4881): P0 — pmp_links backup legs in place and P1 — target schema vetted againstpm-stack/SHARED-SCHEMA-MAP.md(PR docs(pm-stack): shared-schema inventory + schema map, slice 1 of #1744 #4692). - Ordering: depends on phase 1 (orchestration state in Supabase) for the orchestrator's own state inputs, and follows phase 2 (pm stack: centralize all service data as shared tables in one database schema #1744) shared-table conventions.
Done check
backlog-*.mdfiles are generated from DB + GitHub state on demand; hand edits are no longer the source of truth, and a regeneration diff against live state is empty.
Acceptance criteria
- The phase's "Done check" in the body above is met and demonstrated against live state (not just code-merged): the before/after behavior is shown on the running pm stack.
- The migration plan preconditions the body names (P0 — pmp_links backup legs in place; P1 — target schema vetted against
pm-stack/SHARED-SCHEMA-MAP.md) are verified in place before the flip, and the verification is recorded on this issue. - The file-based path keeps working until the DB path is verified; the flip is reversible and no big-bang cutover is required.
Activity
developing-today commentedon Nov 3, 2026
agent-note:v1
state: reviewed (coverage loop-back)
Coverage pass 2026-11-03, loop-back review: this issue was re-reviewed (drift / overlap / already-handled / stale / blockers / ordering) so every open issue in the four repos is reviewed and in the system. Verdict: still live; in the system as task 8573 (triage, queued, priority 40) in repo-health-code-pilot. Labels audited and synced. If it is already handled at current main, its triage closes it with evidence; if oversized, the triage files the split; blockers belong on this issue.
Found via: coverage loop-back 2026-11-03
dezren39 commentedon Nov 3, 2026
Intake review
Fleet monitor intake (Duty 2), 2026-11-03. (Posted on a delayed retry: the original comment attempts were rejected by a GitHub secondary rate limit; the label/body enrichment itself landed during the run.)
pm-stack/SHARED-SCHEMA-MIGRATION-PLAN.md(PR pm-stack: per-service migration plan for the shared schema (#1744 slice 2) #4881) andpm-stack/SHARED-SCHEMA-MAP.md(PR docs(pm-stack): shared-schema inventory + schema map, slice 1 of #1744 #4692).dezren39 commentedon Nov 3, 2026
Triage verdict — task 8573 (lane code-p): still live, correctly blocked on phase ordering — do not start yet
🤖 Agent note · triaged
Verdict: live and accurately scoped, but implementation is blocked by its own stated ordering — this is a sequencing blocker, not a Drew decision and not a defect.
Verified fresh this claim (gh):
effort/l: this is a multi-day migration slice, not a small fix; no single fix task is enqueued from this triage because any implementation claim now would only re-park on the same ordering.Resume condition (named, checkable): start phase 4 implementation when (a) #5005 (phase 1) is closed/done, (b) #1744's shared-table conventions are landed, and (c) the P0/P1 verification notes are posted on this issue. At that point enqueue a
small_fix-series or a dedicated implementation task set for #5007 — one slice at a time, markdown shape unchanged for consumers until the derived view's regeneration diff is empty, per the body's done check.Nothing here needs Drew: the blockers are other fleet work items with their own queue tasks.