Verified

Database MCP Server

SQL database access for AI clients with read-only defaults and row limiting tools.

Data last scanned today · Reviewed today

532 tools★ 5PythonMIT

Overview

Database MCP Server provides structured tools for AI clients to interact with various SQL databases. It supports SQLite, libSQL, MySQL, MariaDB, and Postgres through dedicated packages available for both TypeScript and Python. Each integration exposes a consistent interface comprising two primary tools: execute_sql and search_objects. These are designed with built-in guardrails, including default read-only execution, configurable row caps, and statement timeouts to manage data access safely.

Our verdict

This server carries a high risk rating of 33/100. The project is maintained by a single contributor and has not yet undergone independent editorial review, which introduces potential supply chain and maintenance risks. It is recommended for development environments only, rather than mission-critical infrastructure, until these authentication and security mechanisms are more formally audited.

Supports five major SQL database engines.
Features built-in guardrails for read-only operations.
Includes configurable row caps and statement timeouts.
Zero direct dependencies reduce potential supply-chain surface.

High security risk due to lack of formal audit.
Single contributor presents significant bus-factor risk.
Authentication methods are currently unverified.

Setup

Claude Desktop

{
  "mcpServers": {
    "database-mcp": {
      "args": [
        "-y",
        "database-mcp"
      ],
      "command": "npx"
    }
  }
}

Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.

No read-only mode — running this grants full read/write access.
No built-in authentication — anyone with access to this config can invoke every tool it exposes.

Tools

Tool	Description	Risk
execute_sql	Executes SQL queries against the database.	medium
search_objects	Progressively discloses database schema information including tables, columns, indexes, and foreign keys.	low

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›Which database engines are currently supported?

The server supports SQLite, libSQL, MySQL, MariaDB, and Postgres.

›How are security guardrails implemented?

The server enforces read-only mode via a SQL guard and session-level settings, while also applying default row caps of 1000 and 30-second statement timeouts.

›How do I provide credentials for networked databases?

Credentials should be supplied via environment variables, files containing secrets, or a YAML configuration file passed at launch; they are never accepted via chat prompts.

›What tools are available to the AI client?

Each server provides two tools: execute_sql for running queries and search_objects for exploring schema, tables, columns, indexes, and foreign keys.

›Does the server log sensitive information?

No, secrets are handled as non-printable types, DSNs are sanitized, and a redaction filter is used to prevent secrets from appearing in logs.