postgres=# select * from about;

A public workshop for agentic Postgres hacking.

We use AI agents heavily to explore real problems, inspect the code, prototype fixes, and test the ugly edges. AI accelerates the work; we verify the details and take full responsibility for every result.

Why “Postgres Hacking”?

It works as the community and site name. “Hacking Postgres” remains the session-series title used by PostgresTV. Both point to the same work.

How to contribute

  1. Join the Telegram group and pick an open work line.
  2. Read the patch, thread, and current blockers before writing code.
  3. Use agents aggressively, but verify their work: inspect the relevant code paths, reproduce the problem, add tests, and probe failure modes and observer effects. You own what you submit.
  4. Keep a fork on GitHub or GitLab and use pull or merge requests for convenient pre-review. PostgreSQL does not use either forge as its official development channel.
  5. Submit the patch and discussion to pgsql-hackers. Add it to CommitFest when appropriate so it can be tracked and reviewed; CommitFest complements the mailing-list thread, never replaces it.
~/postgres-hacking/README

$ values

public evidence > recollection

tests > confidence

small patches > grand plans

review is work

agents assist; authors own

$ status

8 open lines

blockers visible

community online

Research note: session summaries and timecodes are AI-generated from timestamped YouTube captions and manually reviewed. Captions can establish what was discussed; links from a session to patches or outcomes still require public evidence. New recordings remain pending until captions exist.