Postgres Pro guides Claude through PostgreSQL performance and administration in five steps: analyze with EXPLAIN (ANALYZE, BUFFERS), design indexes, rewrite queries, set up replication and monitor maintenance through the pg_stat views.
It includes an end-to-end example that finds slow queries in pg_stat_statements, builds a partial index concurrently and confirms the planner uses it, plus SQL for JSONB, dead tuples and replication lag.
What You Get
- A five-step workflow from performance analysis to monitoring
- An end-to-end slow-query example using pg_stat_statements, EXPLAIN, a concurrent partial index and ANALYZE
- SQL patterns for JSONB containment with GIN indexes, dead tuple and bloat monitoring and replication lag
- Reference files for performance, JSONB, extensions, replication and maintenance
- Rules such as CREATE INDEX CONCURRENTLY in production, connection pooling and never disabling autovacuum globally
Common Use Cases
- Finding and fixing the slowest queries on a production database
- Indexing JSONB event payloads for containment queries
- Setting up streaming or logical replication and watching lag
- Tuning autovacuum on high-churn tables
- Choosing and using extensions such as pg_trgm, PostGIS or pgvector
What Makes It Unique
The skill pairs each recommendation with a verification query. Indexes are confirmed with EXPLAIN before and after creation, replication lag is read from pg_stat_replication, and the output template asks for EXPLAIN output with interpretation, rationale for every index and before and after configuration values.
Under The Hood
The skill is instructions and reference markdown that Claude loads by topic; it ships no scripts. Its examples are SQL statements that read pg_stat views, run EXPLAIN ANALYZE, create indexes, run VACUUM and ANALYZE, so when Claude has database access it executes commands that change a live database. Without access it writes the statements for you to run.