MY-MCPSV-POSTGRES

# MY-MCPSV-POSTGRES ## Description Sample source for MCP Server that accesses PostgreSQL using FastAPI and FastMCP. Provides sample source for developers new to MCP Server. ## Usage ### Prepare Tables and Data * Create `inventory` table in PostgreSQL using `create_inventory.sql` * Load `sample_data.csv` ### Prepare API * Install uv ``` powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex" ``` * Create project folder and create python virtual environment ``` cd MY-MCPSV-POSTGRES uv init API-SERVER ``` * Install packages ``` cd API-SERVER uv add fastapi uvicorn uv add "psycopg[binary]" ``` * Place `main.py` in `API-SERVER` folder and update connection string according to your environment. The connection string is set in the `DB_PATH` variable in `main.py`. * Start FastAPI ``` uv run uvicorn main:app --reload ``` ### Prepare MCP Server * Create project folder and create python virtual environment ``` cd MY-MCPSV-POSTGRES uv init MCP-SERVER ``` * Install packages ``` uv add "mcp[cli]" uv add httpx ``` * Place `server.py` in `MCP-SERVER` folder * Start MCP Server ``` uv run mcp dev server.py ``` ### Configure MCP Server on MCP Client * Add to `mcp.json`. For Claude Desktop, use `claude_desktop_config.json` ``` "mcpServers": { "INVENTORY": { "command": "uv", "args": [ "run", "--directory", "C:\\MY-MCPSV-POSTGRES\\MCP-SERVER", "python", "server.py" ] } }, ``` ### Usage from MCP Client ``` Check the inventory of fruits in the inventory management system, and search for products with an inventory of 10 or less. ```