mcp

This MCP server implements the Routine Model Context Protocol server. It is designed to work in tandem with the Routine application and exposes an MCP endpoint over the standard in/out stream used by MCP-compatible clients. To use it, start the server using the recommended command (for example via npx routine-mcp-server or via your MCP client configuration). Once running, you can communicate with it by sending JSON requests to its standard input and reading responses from standard output, following the MCP protocol conventions. The server is intended to be used in environments where Routine is installed and configured; in Claude Desktop setups, you would point your claudedesktop or similar client configuration to the routine MCP server endpoint as demonstrated in the project readme. Tools and capabilities are centered around MCP message handling for routine-related model context interactions, routing requests to the Routine application, and returning MCP-compliant responses.

Prerequisites:

• Node.js (recommended LTS) and npm or yarn installed on your system
• Access to the internet to install the routine-mcp-server package

Installation steps:

1. Install dependencies for local development (optional if using npx):

   • yarn install or
   • npm install

2. Build the project (if you are contributing or running a local build):

   • yarn build

3. Run the MCP server (examples):

   • npx routine-mcp-server
   • Or, install locally and run from the project root: yarn && yarn start

4. If integrating with Claude Desktop or other clients, update your client configuration to point at the server according to the examples in the README (e.g., in claude_desktop_config.json, set the command to the node executable and args to the built dist/index.js when running in development).

Notes and tips:

• The server communicates over stdin/stdout using the MCP protocol; ensure your client is configured to use the same IPC model.
• In development scenarios, you can use yarn start to run the server, which will start the MCP loop and listen for requests.
• Common issues include misconfigured MCP client endpoints or mismatched protocol versions. If you encounter problems, verify that the server process is running and that the client is sending correctly structured MCP messages.
• If you plan to deploy, consider pinning the exact version of routine-mcp-server in your package.json or your deployment scripts to avoid breaking changes.