MCP Weather Server

A weather query MCP server based on FastMCP, providing current weather, future forecasts, Celsius/Fahrenheit conversion, popular city resources, and travel weather advice prompts.

The project includes two entry points:

• server.py: Local stdio mode, suitable for local MCP clients like Claude Desktop, Cursor, and Cherry Studio.

• server_remote.py: Streamable HTTP mode, suitable for deployment to remote hosting platforms like ModelScope.

Features

• get_weather(city: str): Query current weather for a specified city.

• get_forecast(city: str, days: int = 3): Query future weather forecast for a specified city, default is 3 days.

• temperature_convert(value: float, from_unit: str): Convert between Celsius and Fahrenheit.

• weather://cities: Returns a resource list of common cities.

• travel_advisor(destination: str, date: str): Generate a travel weather advice prompt.

Weather data is provided by wttr.in, no additional API tokens required.

Related MCP server: weather-mcp-server

Requirements

• Python >= 3.12

• It is recommended to use uv for dependency management

Install Dependencies

uv sync

If you are not using uv, you can also install dependencies in a virtual environment:

pip install "mcp[cli]>=1.27.0" "httpx>=0.28.1"

Local Execution

stdio mode:

uv run python server.py

MCP client configuration example:

{
  "mcpServers": {
    "weather-server": {
      "command": "uv",
      "args": ["run", "python", "server.py"]
    }
  }
}

Remote Execution

Streamable HTTP mode:

uv run python server_remote.py

Default listening on:

http://0.0.0.0:8000/mcp/

Remote MCP client configuration example:

{
  "mcpServers": {
    "weather-server": {
      "url": "http://127.0.0.1:8000/mcp/"
    }
  }
}

ModelScope Deployment Instructions

When creating or deploying an MCP Server on ModelScope, you can enter the GitHub repository address of this project and use the remote entry file:

python server_remote.py

The service will start in Streamable HTTP mode, listening on 0.0.0.0:8000, with the MCP endpoint at:

/mcp/

If the platform supports custom startup commands, it is recommended to use:

uv run python server_remote.py

Or use it after installing dependencies in the platform environment beforehand:

python server_remote.py

Project Structure

.
├── README.md
├── pyproject.toml
├── server.py
├── server_remote.py
└── uv.lock

License

If you need to release it as open source, please add a License file according to your actual situation.