Weather

You can expose a weather data tool to AI agents through an MCP server. This setup lets an AI agent request current weather for a given location, using a real external weather data source. It enables you to test, chain, and reuse weather queries in prompts and tool-assisted workflows.

How to use

You access the weather tool via an MCP client. The tool is named get_weather and it returns the current weather for a specified location. Use it to let an AI agent ask for weather in cities, regions, or locations you specify. You will provide the location as input, and the server will fetch real-time data from the OpenWeatherMap API. Ensure your client propagates the location to the tool so the agent can obtain up-to-date weather informa­tion during conversations.

How to install

Prerequisites: you need a working environment for running Python and the MCP tooling that drives the server. Ensure you have a shell with access to the uv tooling used to manage the MCP server.

uv venv

source .venv/bin/activate

uv sync

Set up your API key for the weather data source. Obtain an API key from OpenWeatherMap and export it to your environment.

export OPENWEATHER_API_KEY=your_api_key_here

Run the MCP server in development mode to test the weather tool. The following starts the server and loads the weather function. After starting, open the inspector in your browser to test with a pre-filled token.

mcp dev src/mcp_weather.py

Configuration and example MCP client settings

You can declare the MCP server in a client config so your AI agent can discover and use the weather tool. The example below shows how to expose the server as a local, stdio-based tool that runs via uv and points at the weather script.

{
  "mcpServers": {
    "weather": {
      "command": "uv",
      "args": ["run", "[...]", "/[PATH_TO_PROJECT]/src/mcp_weather.py"],
      "env": {
        "OPENWEATHER_API_KEY": "your_api_key_here"
      }
    }
  }
}

Additional MCP server options and third-party examples

You can integrate additional MCP servers to extend capabilities. For example, you can configure a Google Sheets MCP server to enable data creation and modification in Sheets from prompts. The following shows how a Google Sheets MCP server can be declared for a tool available to your agent.

{
  "google-sheets": {
    "command": "uvx",
    "args": ["mcp-google-sheets@latest"],
    "env": {
      "SERVICE_ACCOUNT_PATH": "[/path/to]/keys/service-account-xxx.json",
      "DRIVE_FOLDER_ID": "XXX"
    }
  }
}

Setting up and troubleshooting uv and uvx

If your system cannot find uv or uvx, you may need to create a symbolic link to the binaries so they are in your PATH.

sudo ln -s ~/.local/bin/uv /usr/local/bin/uv
sudo ln -s ~/.local/bin/uvx /usr/local/bin/uvx

Notes and tips

The weather tool uses the OpenWeatherMap API, so keep your API key secure and do not expose it in public prompts or logs. When testing, you can validate the tool by supplying a location such as a city name and verifying that the returned data reflects current weather conditions.

Available tools

get_weather

Get current weather for a location using the OpenWeatherMap API