Production recovery

  • Supabase project found INACTIVE (auto-paused ~Jun 29, down ~3.5 weeks unnoticed). Restored; data intact (56 tables, 15 users, 77 events, 1 waitlist signup).
  • Migration ledger drift repaired (migration repair): local ↔ remote now 0-drift; phantom 20260622120000 beta_access_gate (no statements, no objects) reverted.
  • All 8 Edge Functions redeployed — the restore had kept only compute-tag-affinity.
  • Keep-alive cron added (flow-platform:.github/workflows/supabase-keepalive.yml, commit 826f148) — requires SUPABASE_ANON_KEY repo secret to activate.

Audit (multi-agent, findings adversarially verified)

  • Mobile: CI dead (Flutter 3.16.9 vs Dart ^3.6.1); test suite hung by misplaced integration test; 17 dead dep lines; 2 stale tests. All fixed on fix/ci-test-hygiene (pushed, ready to merge): analyze 0 issues, 108/108 tests, analyze time 312→41 s.
  • Docs: context-map routed API/deployment work to legacy microservices giants; database-schema had 3 confirmed false claims (fixed); auth-screens self-contradiction (fixed); orphans resolved (empty API Reference.md + docs/ deleted; feat-event-series.md + design-system-v2.md committed).
  • Strategy: roadmap.md and feature-tracker.md described the deleted microservices stack — both rewritten in English against verified reality; new target-study with cited market research.
  • IG tool: queue JSON corruption fixed (atomic writes), dead compute_stats wired into digest cover (data-driven copy). Note: tools/ still has no git repo.
  • Stale memory corrected: l10n is 39/43 screens done (not “~30 pending”).

Key numbers surfaced

0 users · 1 waitlist signup · 0 IG posts published · ~63–66k target-age population in RE province.