mcp -taiwan-weather

This MCP server provides access to Taiwan’s Central Weather Bureau data via the Model Context Protocol. The primary tool exposed is get_taiwan_weather_forecast, which returns the 36-hour weather forecast for Taiwan’s counties and cities. To use it, supply locationName (a Taiwan county or city) as input, and you’ll receive the corresponding forecast data from the CWB API (formatted as the MCP response). The server expects an API key (CWA_API_KEY) to authorize requests to the CWB Open Data API. When configured in an MCP-enabled app (such as Claude Desktop or other MCP clients), you can query weather forecasts by referencing the taiwan-weather MCP server and calling get_taiwan_weather_forecast with the locationName parameter, then use the returned JSON (key records) in your downstream prompts or tooling.

Prerequisites:

• Node.js (LTS version) and npm/yarn installed on your system
• Access to a CWA API key (CWA_API_KEY) for the Central Weather Bureau data

Install and run (example using npx):

1. Ensure Node.js and npm are installed.
2. Install and run the MCP server via npx (no local install required):

# Using the provided MCP server package via npx
npx -y @gonetone/mcp-server-taiwan-weather

3. If you need to configure a specific environment, prepare a config snippet (see next step).
4. Alternatively, clone the repo, install dependencies, build, and run locally (if you prefer a local development setup).

# Example local setup (once you have the repository)
git clone https://github.com/GoneTone/mcp-server-taiwan-weather.git
cd mcp-server-taiwan-weather
npm install
npm run build
# Run with your preferred method (node, npm script, or PM2, depending on project setup)

Configuring for MCP clients (example for Claude Desktop):

• Create or edit claude_desktop_config.json to include the taiwan-weather MCP server with the required CWA_API_KEY.

{
  "mcpServers": {
    "taiwan-weather": {
      "command": "npx",
      "args": [
        "-y",
        "@gonetone/mcp-server-taiwan-weather"
      ],
      "env": {
        "CWA_API_KEY": "<您的中央氣象署 API 授權碼>"
      }
    }
  }
}

Tip: If you encounter Windows-specific npx issues, you can switch to using cmd to run npx, or install the MCP server globally and run with node as shown in the repository’s guidance.

Notes and tips:

• You must obtain a CWA API key from the Central Weather Bureau and provide it via the CWA_API_KEY environment variable when configuring the MCP server.
• The primary tool exposed is get_taiwan_weather_forecast, which returns 36-hour forecasts for Taiwan’s counties/cities using the CWB API data (key records).
• When integrating with Claude Desktop or other MCP-compatible apps, place the configuration snippet in the app’s MCP config file.
• If using Windows, consider using the provided cmd-based npx configuration or installing the MCP server globally and running via node, as documented in the README.
• Build and test: you can validate the MCP integration with MCP Inspector to simulate requests and inspect responses.