MCP 服务器

Weather Server

Provides access to weather alerts and forecasts for US locations using the National Weather Service public APIs.

README

weather — Enhanced MCP Weather Server

A feature-rich MCP (Model Context Protocol) server that provides comprehensive weather data using the NWS (National Weather Service) public APIs. Includes input validation, error handling, geocoding, and a unique weather comparison feature.

Quick start

Install uv (macOS / Linux):

curl -LsSf https://astral.sh/uv/install.sh | sh
# restart your terminal

Create project, venv and install deps:

uv init weather
cd weather
uv venv
source .venv/bin/activate
uv add "mcp[cli]" httpx

Run the server:

uv run main.py
# or
uv run weather.py

Available Tools

The server provides the following MCP tools:

1. `get_alerts(state: str)`

Get active weather alerts for a US state.

Input: Two-letter US state code (e.g., "CA", "NY", "TX")
Features: Validates state codes, provides detailed alert information

2. `get_forecast(latitude: float, longitude: float)`

Get detailed weather forecast for a location using coordinates.

Input: Latitude (-90 to 90) and Longitude (-180 to 180)
Features: 10-period forecast (5 days), includes humidity and precipitation probability

3. `get_forecast_by_city(city_name: str)`

Get weather forecast by city name (no coordinates needed!).

Input: City name (e.g., "San Francisco", "New York", "Chicago")
Features: Automatic geocoding, works with city names

4. `get_current_conditions(latitude: float, longitude: float)`

Get real-time current weather conditions.

Input: Latitude and Longitude
Features: Temperature, humidity, wind, pressure, visibility, and more

5. `compare_weather(location1: str, location2: str)`

Compare weather conditions between two locations side-by-side.

Input: Two locations (city names or "lat,lon" format)
Features: Side-by-side comparison, temperature difference analysis, supports both city names and coordinates

Features

✅ Input Validation: Validates state codes and coordinate ranges
✅ Better Error Handling: Specific, helpful error messages
✅ Geocoding Support: Search by city name using Open-Meteo geocoding API
✅ Current Conditions: Real-time weather observations
✅ Enhanced Forecasts: More detailed information including humidity and precipitation
✅ Weather Comparison: Unique feature to compare two locations
✅ Location Detection: Automatically detects and displays city/state names

Files

weather.py — Main MCP server implementation with all tools
main.py — Entry point that imports and runs the server

Notes

Python 3.11+ required (see pyproject.toml)
NWS Coverage: The NWS API only covers US territories. International locations will not work.
Transport: The server uses stdio transport by default
MCP Clients: Use Claude Desktop, mcp CLI, or another MCP client to connect
Geocoding: City name search uses Open-Meteo's free geocoding API (no API key required)

Example Usage

Once connected to an MCP client, you can use the tools like:

# Get alerts for California
get_alerts("CA")

# Get forecast by city name
get_forecast_by_city("San Francisco")

# Get current conditions
get_current_conditions(37.7749, -122.4194)

# Compare weather between two cities
compare_weather("New York", "Los Angeles")