Query Regression Sentinel

A local agent that watches Postgres/MySQL and alerts on query performance regressions

Problem

Database performance issues are reactive: DBAs start manually browsing pg_stats and other internals only after complaints about slowness. An unbounded query can bring down an entire database before anyone notices. A Show HN post on this exact workflow confirms the pain and the lack of proactive tooling.

Solution

A lightweight local agent that periodically samples query plans and execution stats, builds a baseline, and automatically detects regressions (e.g., a query suddenly takes 2x longer or scans 10x more rows). Alerts via Slack/email/webhook with before-and-after comparison and likely causes.

Users

DBAs, backend engineers, and ops at small-to-mid teams, especially SaaS companies without a dedicated DBA.

MVP

Postgres and MySQL connection config; scheduled sampling of pg_stat_statements / performance_schema; simple baseline comparison against historical data; threshold or percentage-based alerting; a diff-style report page or webhook payload.

Revenue

Subscription per monitored database instance (e.g., $10/instance/month) with a 14-day free trial.

Why now

AI-generated code is flooding production with uneven query quality; small teams lack resources for proactive performance management. An indie dev can build an MVP in Go/Python in weeks.