mcp-weather-service

This MCP server implements a simple note-based weather service. It exposes a single resource type: notes that are accessible via a custom note:// URI scheme. Each note has a name, description, and text/plain content type. The server also provides a single prompt: summarize-notes, which generates a summary of all stored notes. You can control the level of detail through an optional style setting (brief or detailed) when requesting the summary. The server includes one tool, add-note, which lets you add a new note by supplying a name and content; the server updates its state and notifies clients of changes to resources. To use these capabilities, run the MCP server and call the available tool to add notes or request a summary of all notes using the summarize-notes prompt. The UI and tooling are designed to work with MCP Inspector for debugging and inspection during development and testing.

Prerequisites:

• Python 3.11+ (or a compatible Python runtime)
• Access to the internet to install Python packages

Development (local testing):

1. Clone the repository and navigate to the project directory: git clone <repository-url> cd weather_service

2. Install the UV toolchain (if not already installed). This example uses UV (Python-based MCP tooling):

   • If you have pipx: pipx install uv
   • If you prefer pip directly: pip install uv

3. Run the server in development mode (directory must contain the weather_service project): uv --directory /path/to/weather_service run weather_service // Example on Windows: uv --directory C:\Prj\weather_service run weather_service

4. To build and publish for distribution via UV, prepare the package: uv sync uv build uv publish // Note: Publish requires PyPI credentials supplied via environment variables or flags: // UV_PUBLISH_TOKEN or --token, UV_PUBLISH_USERNAME/UV_PUBLISH_PASSWORD

Public deployment (using uvx):

1. Ensure the weather_service package is published to your Python package index and accessible by name.
2. Run the server in publish-ready mode: uvx weather_service

Optional testing with MCP Inspector:

• Install MCP Inspector and attach to the running MCP server to debug interactions: npx @modelcontextprotocol/inspector uv --directory C:\Prj\weather_service run weather-service

Notes:

• Replace placeholder paths with your actual project path.
• If you are publishing to PyPI, configure credentials as described above.

Tips and common issues:

• The server uses a single tool (add-note) for creating notes and a single prompt (summarize-notes) for summarizing all notes. Use the note:// URI scheme to access individual notes by name.
• When debugging, MCP Inspector is strongly recommended to visualize requests and responses.
• If you encounter authentication issues during publishing, set UV_PUBLISH_TOKEN or UV_PUBLISH_USERNAME/UV_PUBLISH_PASSWORD as environment variables or pass --token, --username, and --password flags as appropriate.
• For development, ensure the weather_service directory contains the expected package structure that UV can load (e.g., setup.py or pyproject.toml depending on your packaging approach).
• The README mentions a Windows-friendly path format; ensure correct escaping in command lines when using Windows PowerShell or Command Prompt.