Find and diagnose the highest-impact SQL queries hitting a live MySQL database by combining dbs-vector's slow-query corpus (top_impacting_/browse_/search_ MCP tools) with live schema validation through a MySQL MCP such as mysql-mcp-server (explain_query, list_indexes, table_size). Use this whenever the user wants to find slow queries, the most expensive / impacting / heavy / costly queries, "top offenders", what's hammering or overloading the database, missing-index opportunities, query optimization or tuning candidates, lock contention, or example SQL for a given table and why it's slow — even if they never say "dbs-vector", "slow log", or "EXPLAIN". Triggers include "which queries are slow", "what's killing the DB", "find expensive queries on <table>", "index opportunities", "why is <table> slow", "query tuning candidates".
Rewrite a slow or inefficient SQL query for performance when an index alone won't fix it — full-table or per-row aggregates that should be rollups, SELECT * with deep pagination, redundant or accidental joins, ORM-generated CASE/subquery noise, or queries already using the right index but still scanning millions of rows. Because a safe rewrite depends on business meaning the SQL text does NOT carry (what a status code means, whether a column is effectively nullable, which columns the caller truly consumes, whether stale/approximate results are acceptable), this skill INTERVIEWS the domain owner with targeted questions before proposing anything, then validates that the rewrite preserves results and improves the plan. Use when the user wants to rewrite, refactor, simplify, or speed up a specific SQL query, or when find-impacting-queries hands off a "rewrite candidate". Triggers: "rewrite this query", "this query is slow and indexing didn't help", "simplify this SQL", "optimize this SELECT/JOIN", "make this aggr