sebastienb/weather-mcp

Weather MCP Server

A simple Model Control Protocol (MCP) server that allows LLMs to fetch weather data for US zipcodes using the Open-Meteo API.

Overview

This project demonstrates how to build a Model Control Protocol (MCP) server. The MCP protocol allows Large Language Models (LLMs) like Claude to interact with external tools and services.

In this case, the server provides a weather tool that LLMs can use to fetch current weather data for a given US zipcode.

How MCP Works

1. MCP Request Format: The LLM sends a request to the MCP server with the tool it wants to use and the parameters for that tool.

2. Server Processing: The MCP server receives the request, identifies the requested tool, validates the parameters, and calls the appropriate function.

3. Tool Execution: The tool function (in this case, the weather tool) performs its task, such as fetching data from an external API.

4. Response Format: The server formats the response according to the MCP protocol and sends it back to the LLM.

5. LLM Integration: The LLM can then use this response to generate more informed and up-to-date content.

Project Structure

• src/index.js: Main entry point that sets up the Express server and handles MCP requests
• src/tools/weatherTool.js: Implementation of the weather tool using the Open-Meteo API

API Endpoints

MCP Endpoint

• URL: /mcp
• Method: POST
• Body:

  {
    "tools": [
      {
        "type": "weather_get",
        "parameters": {
          "zipcode": "90210"
        }
      }
    ]
  }
  

• Response:

  {
    "outputs": [
      {
        "type": "weather_get",
        "status": "success",
        "data": {
          "zipcode": "90210",
          "location": "Beverly Hills, CA",
          "temperature": {
            "value": 72.3,
            "unit": "°F"
          },
          "humidity": {
            "value": 65,
            "unit": "%"
          },
          "precipitation": {
            "value": 0,
            "unit": "inch"
          },
          "wind": {
            "speed": {
              "value": 5.3,
              "unit": "mph"
            },
            "direction": {
              "value": 270,
              "unit": "°"
            }
          },
          "condition": "Clear sky",
          "timestamp": "2023-11-18T12:00:00Z"
        }
      }
    ]
  }
  

Installation

1. Clone this repository
2. Install dependencies: npm install
3. Start the server: npm start

Usage with LLMs

To use this MCP server with an LLM like Claude, you would need to:

1. Deploy this server somewhere it can be accessed
2. Configure the LLM to use this server as an MCP endpoint
3. The LLM can then make calls to the weather_get tool to get current weather information

License

MIT