Mastering

This repository demonstrates a minimal Model Context Protocol (MCP) server that exposes a single weather tool alongside two client entry points:

• A configuration JSON (src/weather-mcp-server.json) that can be consumed by MCP-aware hosts.
• A LangChain-based Python client (src/langchain_weather_client.py) that invokes the MCP tool via OpenAI function calling.

Setup

1. Create and activate a virtual environment (optional but recommended).

2. Install the project dependencies:

   pip install -e .
   

3. Update the .env file with a valid OPENAI_API_KEY for the LangChain example.

Running the MCP Server

The server exposes a single tool named get_weather_by_city that always returns the same weather string. Launch it with:

python src/weather-mcp-server.py

The server communicates over stdio and is ready to be consumed by any MCP-compatible client.

Using the Sample LangChain Client

With the server available on stdio, run:

python src/langchain_weather_client.py

The script loads the OpenAI API key from .env, asks “I'm going to paris. do I need a raincoat or a winter coat?”, and lets the agent decide when to call the MCP tool before composing a final answer.

JSON Configuration

src/weather-mcp-server.json provides a simple MCP host configuration snippet that launches the server using stdio. Point your MCP-compatible client to this file (or replicate its content) to register the tool.