Ops & Telemetry Board

Purpose: give the team a single, glanceable QA surface before we flip Supabase or feature flags. Pair this with beta-coverage.md and the Phase 0 deploy checklist whenever we prep a build. The board now pulls real data (feature flag config, dual-write plan statuses, Supabase/Convex file timestamps) at build time.

What lives here

1. Feature flag telemetry
   • Tracks status/owner/dependencies for aiContextHints, sessionKinds, dualWriteGuards, etc.
   • Shows which cohorts currently see the flag and whether we have a storage plan.
2. Dual-write checklist
   • Mirrors the plan from /beta/dual-write but adds live status (Healthy/Warning/Blocked).
   • Useful for gating Supabase writes and backfills; if any step is blocked, keep Phase 0 write-protected.
3. Telemetry feeds
   • Supabase migrations, Convex deploys, feature-flag syncs, and snapshot regeneration timestamps.
   • Operates like a mini runbook so we know the last time generate:phase0-seed executed.

Rendered at /beta/ops-telemetry and sourced from apps/stps/src/lib/stps/beta/opsTelemetry.ts.

Usage

1. Update the canonical sources (featureFlags.ts, dualWritePlan.ts, Supabase files). The board reads them automatically via collectOpsTelemetry.
2. Treat status colors as gates:
   • Healthy → ship it.
   • Warning → engineering review needed.
   • Blocked → do not flip production knobs yet.
3. Record run timestamps after each Supabase migration, snapshot regenerate, or flag sync so the rest of the pod can audit the history.

Next steps

• Once dual-write is live, connect this board to actual telemetry (Convex cron or Supabase functions) instead of file timestamps.
• Add alert hooks (email/Slack) when any feed flips to “Blocked”.
• Extend the board with error budgets (e.g., number of failed Supabase migrations per week) once we have enough data.