天气工具服务器

MCP Weather Tool Server

This project is a demonstration of the mcp library to create a tool server that provides weather-related information. It also includes other example tools and resources.

This is intended as a hands-on learning project for understanding how an MCP server works.

Quick Start

1. Start the MCP server by running:

./run.sh

2. Create a .vscode/mcp.json file with the following content:

{
	"servers": {
		"weather-http": {
			"url": "http://localhost:8000/mcp",
			"type": "http"
		}
	},
	"inputs": []
}

This will start the weather_streamable_http server that you can connect to via the MCP protocol.

Features

This server exposes several tools and resources:

Weather Tools

• get_alerts(state: str): Get weather alerts for a US state.
  • state: Two-letter US state code (e.g., CA, NY).
• get_forecast(latitude: float, longitude: float): Get the weather forecast for a specific location.
  • latitude: The latitude of the location.
  • longitude: The longitude of the location.

Other Tools

• add_2_numbers(a: int, b: int): Adds two numbers together.
• greet_user_formal_tool(name: str): Greets a user formally.
• greet_user_street_style_tool(name: str): Greets a user in a "street" style.

Prompts

• greet_user_prompt(name: str): A prompt that will greet the user '''Laurent''' formally, and all other users in a "street" style.

Resources

• resource://weather_glossary/: Retrieves a full glossary of weather-related terms.
• resource://weather_glossary_term/{word}: Retrieves the definition of a specific term from the weather glossary.

Setup and Installation

1. Install the required Python packages:

   复制

   pip install -r requirements.txt
   

Running the Server

This project includes three different implementations for running the MCP server.

1. Standard I/O (stdio)

This runs the server over standard input/output.

python weather.py

2. Server-Sent Events (SSE)

This runs the server using Server-Sent Events.

python weather_sse.py

3. Streamable HTTP Server

This runs the server as a streamable HTTP application using uvicorn. This is the most feature-rich example, including the glossary resources and extra tools.

You can run it using the provided shell script:

./run.sh

This will start the server on http://localhost:8000.