postgresql
PostgreSQL
Build correct, maintainable PostgreSQL data access before adding complexity. The ordinary target is roughly 100 MB–1 GB of data; concurrency, access patterns, and recovery needs matter more than a size label.
Use sql for shared relational semantics and sql-performance for the measurement and optimization loop. Language database skills own driver APIs and resource cleanup. This skill owns PostgreSQL behavior and changes to its schema and data access.
Establish the database contract
Read the existing schema, migrations, query callers, and operational constraints. Determine:
- deployed server major, extensions, driver/ORM, migration runner, and pool mode;
- domain invariants, transaction boundaries, tenant/owner boundaries, and expected results;
- relevant row counts, data distribution, table/index sizes, read/write mix, and concurrency;
- target environment, connection role, available tools, recovery route, and allowed interruption.
Use an available SQL client or application connection; do not assume psql, containers, extensions, or administrator access exist. SHOW server_version identifies the server; a local client version does not. For consequential syntax or operational behavior, open the official documentation for that deployed major. The /current/ links below are discovery links, not a minimum supported version.