Two things are on offer and they solve different halves of the problem: search grounds an answer in the manual for the version you are actually on, and the skills carry the opinions a manual does not — which constraints and indexes belong on a table before it ships.
A hosted MCP server run by TigerData with semantic and keyword search over official database documentation, paired with a set of opinionated PostgreSQL skills an agent can pull in while it writes schema.
- `search_docs` — one search tool across three corpora, supporting both semantic vector similarity and BM25 keyword search: `postgres` for the official PostgreSQL manual scoped by version, `tiger` for TimescaleDB and its ecosystem, and `postgis` for the spatial extension
- `view_skill` — curated PostgreSQL best-practice skills covering schema design, indexing strategies, data types, data integrity and constraints, naming conventions, performance tuning and modern PostgreSQL features
Nothing to install for the search half: point your client at the public endpoint `https://mcp.tigerdata.com/docs` over HTTP. The skills install separately with `npx skills add timescale/pg-aiguide --skill postgres`, which works with Claude Code, Cursor, Codex, Gemini CLI and VS Code. Claude Code users can instead take both at once through the plugin marketplace: `claude plugin marketplace add timescale/pg-aiguide` then `claude plugin install pg@aiguide`.
One command — npx skills add timescale/pg-aiguide --skill postgres