PlanSleuth

PlanSleuth is a web app (with an optional lightweight desktop agent) that continuously captures query plans and runtime stats from Postgres and MySQL, then flags regressions the moment they appear. It focuses on the ugly, real-world problem: the same query gets slower after a deploy, data growth, or a stats change, and nobody notices until dashboards time out. The app correlates plan diffs, parameter patterns, table bloat, index usage, and lock/wait events to produce a short, actionable “why it got slower” report. It also generates safe, reviewable recommendations (index candidates, query rewrites, stats/analyze cadence, and config hints) with rollback guidance. This is a combination traditional + AI app: traditional collection/analysis plus AI to summarize root cause and propose fixes. It’s not a magic auto-tuner; it’s a pragmatic regression detector and triage tool.