15d • General discussion
Feedback from the SQL performance factory: first full run
We pointed an agent pipeline (intake, spec, build, gate, review, approve) at our whole SQL audit backlog, 45 findings in one go. We also adapted the factory to our way of working as we went:
• Scope control: we agreed up front to skip the DBA and infra items and run only the findings that touch our code, two jobs at a time.
• Ticket hygiene: we held rejected and duplicate findings back so they didn't each open a Jira ticket.
• Ticket and branch reuse: the factory can now continue a ticket's existing branch. When that branch turned out to carry other tickets' commits, we cut a clean one from develop with only the new work, and added a command to repoint the job.
• Rollback rule: the DDL reviewer pointed out that our deployment runs every .sql file, so rollback copies there were dangerous. We kept a single canonical rollback and taught the lint to accept it.
• Human decisions: the pipeline stops and asks whenever evidence is thin, a ticket is shared, or the risk is high.
The result: a strong filter and reviewer that is shaped around our process, with a human decision before anything ships.
2
1 comment
Feedback from the SQL performance factory: first full run
powered by
skool.com/zazencodes-agentic-coding-club-7823
Learn how to use coding agents like the top 1%. Become the AI expert on your team. Welcome to the official ZazenCodes community.
Suggested communities
Powered by