Weather MCP Server
This project implements a Model Context Protocol (MCP) server that provides real-time weather forecasts and alerts. It is built using the Python MCP SDK and the National Weather Service (NWS) API. The server fetches data (forecasts, alerts, observations, etc.) from the NWS API and exposes them as MCP tools that AI assistants (e.g. Claude) can call. In short, it allows you to ask Claude questions like “What are the weather alerts in CA?” or “What’s the forecast for 37.77, -122.42?”, and the server will return up-to-date information.
Prerequisites
- Python 3.11 or higher. The Python MCP SDK requires a recent Python version.
- uv package manager. This lightweight tool manages Python dependencies and can run scripts. (Install it via
curl -LsSf https://astral.sh/uv/install.sh | shon macOS/Linux or using PowerShell on Windows, or viapipx install uv.) - MCP Python SDK and HTTP library. We'll install these with
uvbelow.
Setup / Installation
-
Clone the repository (or download the code):
git clone https://github.com/Danii2020/weather-mcp.git cd weather -
Install dependencies with
uv. In the project directory, run:uv add mcp[cli] httpxThis installs the Python MCP SDK (
mcp[cli]) and an HTTP client (httpx) used by the server. -
(Optional) If your project includes a
pyproject.tomlor other dependencies, you can install them similarly. But the above command covers the core libraries needed.
Running the Server
Start the weather MCP server by running:
uv run weather.py
This will launch the server (using uv to manage the environment). The terminal will print status messages. Keep this process running to serve requests.
Note: If you run into issues, make sure you have activated the correct Python environment and that
uvis in your PATH.
Configuring Claude for Desktop
To let Claude for Desktop use this weather server, you must add it as an MCP server in Claude’s config.
-
Platform support: Claude Desktop is available for macOS and Windows only (Linux is not supported).
-
Config file location: Find or create the file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
- macOS:
-
Open
claude_desktop_config.jsonin a text editor. Add an entry under"mcpServers"for the weather tool. For example:{ "mcpServers": { "weather": { "command": "/full/path/to/uv", "args": ["run", "weather.py"], "cwd": "/full/path/to/weather" } } }- Replace
/full/path/to/uvwith the absolute path of theuvexecutable. You can find this by runningwhich uvon macOS or Linux, orwhere uvon Windows. - Replace
/full/path/to/weatherwith the full path to theweatherproject directory on your machine.
- Replace
-
Save the file and restart Claude for Desktop. In Claude, open the Developer settings and ensure the “weather” server appears under available MCP servers. You can now select it in your conversation.
Usage Examples
Once the server is running and Claude is configured, you can ask Claude to use the weather tools. For example:
- Weather Alerts: Ask “What are the current weather alerts in CA?” (the server’s
get_alertstool will fetch alerts from NWS). - Forecast: Ask “What’s the 5-day forecast for latitude 47.6, longitude -122.3?” (the server’s
get_forecasttool will retrieve the forecast).
Claude will display the results returned by the server. You can experiment with different state codes or coordinates as needed.
Video Tutorial
For a step-by-step walkthrough, watch the author’s YouTube tutorial “Learn MCP from Scratch and Build an MCP Server with Python!” at https://youtu.be/Pu5Q2dDwR9w. The video shows how to set up uv, code the server, and connect it to Claude Desktop.