MCP PostgreSQL Demo

What is MCP PostgreSQL Demo?

A FastMCP server that enables LLMs to connect and interact with PostgreSQL databases. It demonstrates how to use the Model Context Protocol (MCP) to allow Language Models to query and explore database schemas and tables.

How to use MCP PostgreSQL Demo?

Install Python 3.12+, PostgreSQL, and UV. Clone the repository, create a virtual environment, install dependencies with uv sync, and configure environment variables. Uncomment the run function in src/main.py and start the server with python -m src.main. Alternatively, use Docker with the provided Dockerfile. For client applications, add the MCP configuration to the client’s MCP config file (e.g., .cursor/mcp.json).

Key features of MCP PostgreSQL Demo

• Schema Exploration: Retrieve metadata about database schemas
• Table Inspection: Get detailed information about table structures
• Database Querying: Execute SQL SELECT queries against the database
• YAML Formatting: Results returned in YAML format for LLM consumption
• Predefined prompts for schema, table descriptions, and data queries

Use cases of MCP PostgreSQL Demo

• Allow LLMs to explore database schemas without direct SQL access
• Enable LLMs to query table data using natural language via prompts
• Provide structured metadata to LLMs for context-aware responses
• Demonstrate MCP integration with PostgreSQL for AI assistants

FAQ from MCP PostgreSQL Demo

What are the prerequisites for running the server?

Python 3.12 or higher, a PostgreSQL database, and UV package manager (recommended).

How do I configure the database connection?

Set environment variables: DB_HOST, DB_PORT, DB_USER, DB_PASSWORD, and DB_NAME. Default values are provided in .env.example.

Can I execute any SQL query?

Only SELECT queries are allowed via the query_database tool; no modifications are supported.

How are query results formatted?

Results are returned in YAML format for easy parsing by LLMs.

Can I run this server in Docker?

Yes, build the image with docker build -t mcp-demo . and run with docker run --env-file .env.docker -p 8000:8000 mcp-demo.