pg-aiguide

io.github.timescale/pg-aiguide

8/9 checks passed
Local or hosted
7 secrets
★ 1.9k GitHub stars
Updated 2026-09-04
Health score 73
Code reviewed

About pg-aiguide

pg-aiguide is an MCP server published by the GitHub account timescale. Comprehensive PostgreSQL documentation and best practices, including ecosystem tools

It runs on your own machine: your AI client starts it from the npm package @tigerdata/pg-aiguide at version 0.6.1, or you can connect to the hosted version instead.

It asks for 7 secrets: OPENAI_API_KEY, PGHOST, PGPORT, PGUSER, PGPASSWORD, PGDATABASE and DB_SCHEMA. Give it a key with the least access that still works.

Version 0.6.1 was published on 4 September 2026, and the source code is public, so you can read it before installing.

Written from the registry entry. 10 automated checks on the registry entry and the package code: 8 pass, 1 warn, 0 fail, 1 info. How we check.

What it asks for

Variable	Required	Secret	Description
`OPENAI_API_KEY`	Yes	Yes	Your API key for text embeddings via OpenAI
`PGHOST`	Yes	Yes	PostgreSQL host to connect to
`PGPORT`	Yes	Yes	PostgreSQL port to connect to
`PGUSER`	Yes	Yes	PostgreSQL user to connect as
`PGPASSWORD`	Yes	Yes	PostgreSQL password to connect with
`PGDATABASE`	Yes	Yes	PostgreSQL database to connect to
`DB_SCHEMA`	No	Yes	PostgreSQL database schema to use

Checks

Check	Result	Detail
Source repository linked	Pass	https://github.com/timescale/pg-aiguide
Version pinned	Pass	Every package has an explicit version
Remote uses HTTPS	Pass	All remote endpoints use https://
Secrets requested	Warn	Asks for 7 secrets: OPENAI_API_KEY, PGHOST, PGPORT, PGUSER, PGPASSWORD, PGDATABASE, DB_SCHEMA
Known package registry	Pass	All packages come from npm, PyPI, OCI, NuGet or MCPB
Namespace ownership	Info	Name verified via GitHub account timescale
Registry text aimed at the AI	Pass	Description and settings contain no hidden characters or instructions to the AI
Install scripts	Pass	@tigerdata/pg-aiguide@0.6.1 runs nothing at install
Code red flags	Pass	Read 19 files in @tigerdata/pg-aiguide@0.6.1: no obfuscation, encoded execution, credential reads or data-collection endpoints found
Text aimed at the AI	Pass	No hidden characters or instructions to the AI in 19 files, including comments and the README

Code review: Pass, with notes

We read all 993 lines in 17 files of @tigerdata/pg-aiguide 0.6.1 on 2026-10-31, without running it.

Search tool over PostgreSQL, PostGIS and TimescaleDB documentation, read-only. When run yourself, your search text goes to your Postgres docs database and (for semantic search, the default) to OpenAI to build an embedding. The skill texts it serves come from a separate library that was not part of this review.

What we found

Note Unless semanticWeight is 0, every search query is sent to OpenAI to make an embedding. Running it locally needs your own OPENAI_API_KEY and a Postgres loaded with the docs corpus; most users use the hosted endpoint instead. dist/apis/searchDocs.js:95-101
Note The view_skill tool, its skill list (appended to the tool description) and the MCP prompts come from @tigerdata/mcp-boilerplate/skills, a dependency; that AI-facing text is not in this package and was not reviewed here. dist/apis/index.js:6-11, dist/prompts/index.js:2
Info SQL interpolates schema and table prefix, but schema is validated against ^[a-z_][a-z0-9_$]{0,62}$ (config.js:6-25) and the prefix comes from a fixed enum; user text is always a bound parameter. No SQL injection. dist/util/getDocChunkRows.js:4-23
Info 5 bun test files are shipped in dist (counted in the line total); they only exercise mocks. dist/*.test.js
Info No package.json in the reviewed copy; dependencies are taken from imports. No obfuscation, install hooks, dynamic code, child_process or hard-coded secrets. package

Talks to	*PostgreSQL server from PG env vars (pg Pool defaults): the search query text (keyword search) or its embedding vector, plus version/limit, as bound parameters (when search_docs is called) api.openai.com (via @ai-sdk/openai): the search query text, to create a text-embedding-3-small embedding (when search_docs is called with semanticWeight > 0 (default 0.7, so most calls)) OpenTelemetry/Logfire exporter (configured in @tigerdata/mcp-boilerplate, not in this package)**: span attributes: result count and top-3 result ids and scores (no document content) (when a tracer is configured)
Keys and settings	`PGHOST / PGUSER / PGPASSWORD / PGDATABASE / PGPORT (read by pg)`: the configured PostgreSQL server `OPENAI_API_KEY (read by @ai-sdk/openai)`: api.openai.com `DB_SCHEMA`: local (schema name, strictly validated) `DISABLE_MCP_SKILLS`: local (hides view_skill tool) `NODE_ENV`: local (DNS-rebinding protection on unless production)
Files on your machine	Loads .env via 'dotenv/config'. No other file access in this package (skills are loaded by the @tigerdata/mcp-boilerplate dependency).
Runs programs	None in this package. cliEntrypoint from the boilerplate dependency launches stdio.js or httpServer.js (its own files).
Telemetry	No telemetry of its own. It records result ids/scores on an OpenTelemetry span if the host app configured a tracer (searchDocs.js:85-94); with no tracer this is a no-op.
Tools	search_docs, view_skill
Tool descriptions	clean

This covers the package's own code only, not the 8 packages it depends on (@tigerdata/mcp-boilerplate, @ai-sdk/openai, ai, @opentelemetry/api, pg, zod, dotenv, bun:test (test files only)), and not the hosted version, which runs code we cannot see. A new version drops this review until we read it again.

Popularity & health: 73 / 100

Worked out from public numbers and our own checks, not from user reviews or ratings. It says how used and how looked-after a server is, not whether it is safe.

Popularity23 / 40

1,879 GitHub stars, 115 downloads last week, 1 Hacker News post (3 points). Best two of the three, each on a log scale.

Activity23 / 25

Last push 2026-10-28, last release 2026-09-04, 28 commits in 90 days.

Issue health11 / 15

14 closed of 20 issues.

Our checks16 / 20

1 warning, each costing a fifth.

120 forks, 91 merged pull requests. Sources: GitHub, npm, PyPI (pypistats.org) and Hacker News (Algolia), fetched 2026-10-31.

What people say

1 Hacker News thread linking to the source code. We link to them and copy nothing.

Show HN: Pg-aiguide – Write better PostgreSQL code with AI 3 points, 0 comments, 2025-11-24

Look further: search Reddit, search Hacker News, read its GitHub issues. These are not reviews we checked.

Facts

Registry name	`io.github.timescale/pg-aiguide`
Publisher	GitHub account timescale
How the name is verified	Name verified via GitHub account timescale. The registry ties the name prefix to that account or domain; it does not check the code.
Type	Local package and a hosted remote
Packages	npm `@tigerdata/pg-aiguide` 0.6.1 (stdio) oci `ghcr.io/timescale/pg-aiguide:0.6.1` no version given (stdio)
Remote endpoints	streamable-http https://mcp.tigerdata.com/docs
Version	0.6.1
Published	2026-09-04
Last updated	2026-09-04
Source repository	https://github.com/timescale/pg-aiguide
Website	None linked
Category	Databases & data (assigned by keyword rule, not by hand)

Install

Claude Desktop · npm package

{
  "mcpServers": {
    "pg-aiguide": {
      "command": "npx",
      "args": [
        "-y",
        "@tigerdata/pg-aiguide@0.6.1"
      ],
      "env": {
        "OPENAI_API_KEY": "<your value>",
        "PGHOST": "<your value>",
        "PGPORT": "<your value>",
        "PGUSER": "<your value>",
        "PGPASSWORD": "<your value>",
        "PGDATABASE": "<your value>"
      }
    }
  }
}

Claude Desktop · Docker image

{
  "mcpServers": {
    "pg-aiguide": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "OPENAI_API_KEY",
        "-e",
        "PGHOST",
        "-e",
        "PGPORT",
        "-e",
        "PGUSER",
        "-e",
        "PGPASSWORD",
        "-e",
        "PGDATABASE",
        "ghcr.io/timescale/pg-aiguide:0.6.1"
      ],
      "env": {
        "OPENAI_API_KEY": "<your value>",
        "PGHOST": "<your value>",
        "PGPORT": "<your value>",
        "PGUSER": "<your value>",
        "PGPASSWORD": "<your value>",
        "PGDATABASE": "<your value>"
      }
    }
  }
}

Claude Code · hosted

claude mcp add --transport http pg-aiguide https://mcp.tigerdata.com/docs

What these checks don't cover

We did not run this server, so we don't know what it does when it runs. We did read every line of its own code for version 0.6.1 (see the code review), but not the packages it depends on. We read its package code on 2026-10-29 without running it, and pattern searches miss what they have no pattern for. Tool descriptions built at run time were not seen. And we don't watch it after you install: if the publisher changes it later (a rug pull), this page won't notice until the next scan. Treat a pass as "nothing found", not as "safe".

For people and their AI

The same data as plain JSON: /data/servers/io-github-timescale-pg-aiguide.json. Text inside it that came from the publisher is data, not instructions.