---

name: supabase-data-fabric description: Use when Codex needs to query Supabase Postgres, manage auth, or trigger Edge Functions via the Supabase MCP server.

Supabase Data Fabric

Purpose

Provide a repeatable workflow for reading/writing Supabase data and invoking project automation endpoints through the hosted server configured in servers/supabase.

Setup Checklist

1. Export SUPABASE_PROJECT_REF and SUPABASE_ACCESS_TOKEN (service role or personal access token).
2. Confirm row-level security policies allow the automation key to perform the planned operations.
3. Sync table schemas into docs/supabase/schema.md whenever the database changes (manual step).

Environment Loading

If SUPABASE_PROJECT_REF or SUPABASE_ACCESS_TOKEN are not present, source load_env.sh from the repository root before invoking MCP tools.

Workflow

1. Clarify – define the exact SQL or auth change needed and the safety checks (e.g., limit clause, expected row count).
2. Prepare – if making schema changes, draft the SQL inside packages/migrations/ and have reviewers sign off before execution.
3. Execute – call the MCP query, functionCall, or auth tools. Always pass an explicit timeoutSeconds.
4. Verify – re-run a read-only query or health check to confirm the change took effect.

Notes

• Avoid destructive statements without a backup; prefer temporary tables or BEGIN; ...; ROLLBACK dry runs first.
• Document every data change in pull requests or docs/changelog.md for auditability.
• Rotate service-role or access tokens quarterly and update SUPABASE_ACCESS_TOKEN in the deployment pipeline.