Weather MCP Server

A demo Model Context Protocol server that gives AI assistants (Claude, Copilot, Cursor, etc.) access to live US weather data via the free National Weather Service API.

Tools exposed

No API key required — NWS is completely free.

Local development

1. Install uv

curl -LsSf https://astral.sh/uv/install.sh | sh

2. Install dependencies

uv pip install "mcp[cli]>=1.0.0" "httpx>=0.27.0"

3. Run with MCP Inspector (for testing)

uv run mcp dev weather_server.py

Opens a browser UI to call your tools manually.

4. Connect to Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "weather": {
      "command": "python",
      "args": ["/absolute/path/to/weather_server.py"]
    }
  }
}

Restart Claude Desktop. You'll see a 🔨 tools icon in chat.

Deploy to Fly.io

1. Install flyctl

curl -L https://fly.io/install.sh | sh

2. Sign up / log in

fly auth signup   # or: fly auth login

3. Pick a unique app name

Edit fly.toml and change app = "weather-mcp-demo" to something unique.

4. Launch

fly launch --no-deploy   # reads fly.toml, provisions the app
fly deploy               # builds Docker image and deploys

Your server will be live at:

https://<your-app-name>.fly.dev/mcp

5. Connect Claude Desktop to the hosted server

{
  "mcpServers": {
    "weather": {
      "url": "https://<your-app-name>.fly.dev/mcp"
    }
  }
}

Useful Fly commands

fly logs          # tail live logs
fly status        # check machine health
fly deploy        # redeploy after code changes
fly scale count 1 # ensure at least 1 machine running

Example prompts to try

• "What's the weather forecast for New York City?" → lat: 40.71, lon: -74.00

• "Any weather alerts in Texas right now?"

• "Give me the hourly forecast for Miami." → lat: 25.77, lon: -80.19