MCP Learning

A simple Python MCP (Model Context Protocol) server built with FastMCP, created while learning how MCP servers work.

What this project does

This server exposes two tools that can be called through an MCP client (like the MCP Inspector):

• get_weather(city) — returns a sample weather response for a given city

• add_numbers(a, b) — returns the sum of two numbers

Related MCP server: Banana

Tech used

• Python

• FastMCP (mcp.server.fastmcp)

• MCP Inspector for testing

How to run

1. Install dependencies:

   pip install "mcp[cli]"

2. Run the server directly:

   py server.py

3. Or launch it through the Inspector (recommended for testing):

   npx @modelcontextprotocol/inspector py server.py

4. In the Inspector UI, click Connect, go to the Tools tab, click List Tools, and try out get_weather or add_numbers.

What I learned

• Setting up a basic MCP server using FastMCP

• Fixing a Windows PATH issue (python not recognized → use py instead)

• Connecting a Node.js-based Inspector to a Python server as a subprocess

• Debugging simple typos in tool decorators (@mcp.tool())

Next steps

• Add more tools

• Explore MCP resources and prompts

• Connect this server to a real MCP client