SQLite MCP Server

by Anthropic (MCP reference servers)

Reference Database

A reference MCP server for local SQLite databases — run queries, inspect table schemas, and build up analysis against a single file with no database server to run.

Add to Claude Code

claude mcp add sqlite -- uvx mcp-server-sqlite --db-path /path/to/database.db

Source: https://github.com/modelcontextprotocol/servers/tree/main/src/sqlite

About

The SQLite MCP server points at a local .db file and gives an agent tools to list tables, describe schemas, run read queries, and execute write queries against it — useful for prototyping, local analytics, or small embedded datasets without standing up a full database server.

It also includes a lightweight "memo" concept in the original reference design, letting accumulated insights from repeated queries be appended back into a running note — a simple pattern for building up analysis over a session.

Key features

  • List tables and describe schemas for a local SQLite file
  • Run both read and write queries
  • No server process required — works directly against a .db file
  • Good fit for prototyping and local data analysis

Use cases

  • Let an agent explore and query a local SQLite dataset during analysis
  • Prototype a data model before committing to a full database server
  • Give a small internal tool or script queryable memory in a single file

Available tools

read_query

Executes a read-only SQL query against the SQLite database.

write_query

Executes an INSERT, UPDATE, or DELETE statement.

list_tables

Lists all tables in the connected database.

describe_table

Returns column names and types for a given table.

Frequently asked questions

Does it need a running database server?

No — SQLite is a single-file database, so the MCP server reads and writes that file directly with no separate server process.

Is it suitable for production workloads?

SQLite itself is fine for many small-to-medium workloads, but for concurrent, multi-user production systems a server-based database like Postgres and its dedicated MCP server is the better fit.