pgblame links slow Postgres queries to the deploy that caused them

pgblame helps a Supabase and Next.js app owner find which deploy made Postgres queries slow. Normal pg_stat_statements can show which queries are slow, but it does not show when the slowdown started or which code change caused it. pgblame runs as a small beside the app and checks pg_stat_statements every minute.

A Vercel or Railway , or a call from CI, lets it match deploy times with query speed changes. If a query changes from 12 milliseconds to 280 milliseconds after a deploy, pgblame points to the related commit. The agent is and read-only, so real app data stays on the server; it only reads query .

The free tier covers 1 database, 7 days of history, and does not require a card. It is still early, with testing feedback especially wanted from Supabase and Neon users.

Key points

  • pgblame connects slow Postgres queries to the deploy and commit that likely caused them.
  • It runs beside the app as a and checks pg_stat_statements every minute.
  • It can receive deploy events from Vercel, Railway, or CI.
  • The agent is and read-only, and it reads query instead of real app data.
  • The free tier includes 1 database, 7 days of history, and no card requirement.
Read original