simple-psql

This MCP server exposes a PostgreSQL-backed set of tools, resources, and prompts that you can access through the MCP Inspector or an integrator AI. The Tools section includes execute_query for running SQL against your database and test_connection to verify connectivity. Resources provide db://tables (lists all tables), db://tables/{table_name} (schema details for a specific table), and db://schema (complete schema information for all tables). Prompts include query generation templates and analytical query builders that help construct SQL queries from natural language intents. To use it, start the server via the Inspector (uv) and connect your AI tool to the configured DSN and schema. The Inspector UI lets you interactively test commands, view results, and see how the MCP server responds to natural language inputs transformed into tool calls.

Prerequisites:\n- Python 3.8+\n- uv (Python package manager/installer)\n- Node.js with npx (for the MCP Inspector)\n- PostgreSQL database you can connect to\n\n1. Set up a Python virtual environment and install dependencies:\nbash\n# Create a virtual environment with uv\nuv venv\n\n# Activate the virtual environment\nsource .venv/bin/activate # On Windows: .venv\Scripts\activate\n\n# Install dependencies\nuv pip install -r requirements.txt\n\n\n2. Start the MCP server via the Inspector (as shown in the README):\nbash\n# Replace with your actual database credentials in DSN/SCHEMA within the config file or environment\nnpx @modelcontextprotocol/inspector uv --directory . run postgres\n\nThis will launch the MCP Inspector interface in your browser where you can connect to the server and test Tools, Resources, and Prompts.\n\n3. Create or update the MCP configuration file (see example in the README) to point to your uv-based server.\njson\n{\n "mcpServers": {\n "postgres": {\n "command": "uv",\n "args": [\n "--directory",\n "/path/to/simple-psql-mcp",\n "run",\n "postgres"\n ],\n "env": {\n "DSN": "postgresql://username:password@localhost:5432/my-db",\n "SCHEMA": "public"\n }\n }\n }\n}\n\n\n4. Optional: If you use the included example database, follow the README’s steps to spin it up and connect via the inspector.\n

Tips and common considerations:\n- Ensure DSN and SCHEMA in the MCP config reflect your actual database connection details. Treat DSN like a secret; do not commit it to public repos.\n- The server performs a basic safety check (e.g., only allows SELECT-like queries via execute_query by default); evaluate whether you need stronger safeguards for production.\n- The Inspector UI is a convenient way to validate commands before integrating with an AI tool.\n- If you modify the SQL templates or prompts, keep them aligned with the actual database schema to improve accuracy of generated queries.\n- If you encounter connection issues, verify network access to the PostgreSQL host, correct port, user permissions, and that the database accepts connections from the environment where the MCP server runs.\n