mcps
P S M R S

Recipe

Query any database with Claude (no SQL reflex needed)

Point Claude at Postgres, MongoDB, Redis, SQLite or a managed cloud DB and ask questions in plain language.

for Founders and analysts who want answers without writing SQL

Hand-checked recipe · curated by Ilya Gindin

10 minutes

to run

beginner

difficulty

5

connectors

What you'll need

PostgreSQL Read-only-friendly access to a Postgres database — schema, rows, aggregates.
Optional
Supabase Managed Postgres + auth + storage when your data lives in Supabase.
Optional
MongoDB Document queries against a local or Atlas MongoDB.
Optional
Redis Key/value and cache inspection for a local or cloud Redis.
Optional
SQLite Zero-setup local SQLite file — great for prototypes and analysis.
Optional

This recipe runs in 3 steps

  1. 1

    Connect your database server

    Say to your AI

    Install the MCP server for my database (Postgres / Mongo / Redis / SQLite / Supabase) and confirm you can see the schema.

    You'll get: Claude lists the tables/collections it can see.

  2. 2

    Understand the shape

    Say to your AI

    Describe my schema: tables, key columns, and how they relate. Flag anything that looks off.

    You'll get: A short map of the schema plus any obvious modeling smells.

  3. 3

    Ask a real question

    Say to your AI

    How many active users signed up in the last 30 days, broken down by week? Write the query, run it, and show the result.

    You'll get: The generated query, the result table, and a one-line read of the trend.

You're done when

You get answers from your data by asking, while still seeing the exact query Claude ran — so you can trust and reuse it.

Why this workflow exists

Most database questions are one-offs: a count, a breakdown, a “did this migration actually work”. Writing the SQL by hand for each is slow, and a raw chatbot will hallucinate column names. A database MCP server lets Claude read your real schema first, then write a query that actually runs — and it always shows you the query, so nothing happens behind your back.

Pick the server that matches where your data lives. The steps are the same whether it’s a local SQLite file or a managed cloud Postgres.

People ask their AI

“query my database with Claude”“what MCP for postgres / mongodb / redis”“explore my database in plain language”

Related

data analysis with supabase postgres