An MCP (Model Context Protocol) server that assesses weather-related danger for driving routes. Use it with Claude Code, Claude Desktop, or any MCP-compatible client.
Combined tool that handles the full workflow in a single call:
- Takes origin, destination, and optional departure/arrival time
- Derives the route and gets waypoints via Google Maps API
- Fetches forecast weather for each waypoint at its expected arrival time via Open-Meteo API
- Computes danger scores for each point
- Returns overall assessment with status (SAFE, MODERATE, HAZARDOUS, EXTREME)
Example: "Compute the danger of traveling from Grayson, GA to Dahlonega, GA on January 23, 2026, leaving at 07:00 AM"
Takes origin/destination cities and optional departure/arrival times. Returns a list of (lat, long) waypoints along the route.
- Python 3.12+
- Google Maps API key (for geocoding and routing)
Set the API key in your environment:
export GOOGLE_MAPS_API_KEY=your_google_maps_api_keyYou'll need to enable these APIs in Google Cloud Console:
- Geocoding API
- Routes API
# Install dependencies with uv
uv sync
# Install dev dependencies (for testing)
uv sync --extra devuv run pytestAdd to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"safe-travels": {
"command": "uv",
"args": ["run", "--directory", "/path/to/safe-travels", "python", "server.py"],
"env": {
"GOOGLE_MAPS_API_KEY": "your_key_here"
}
}
}
}Note: Replace /path/to/safe-travels with the actual path to this project.
Then restart Claude Desktop and ask questions like:
- "What's the danger level for driving from Denver to Boulder tomorrow morning?"
- "Assess the route from Atlanta to Savannah"