daymon

Daymon provides a local, zero-cloud automation platform that runs Claude-powered tasks on your own machine. The MCP (Model Context Protocol) server for Daymon is exposed as a standalone Node.js sidecar that handles cron scheduling, file watchers, and background tasks, enabling Claude interactions to be executed locally without external API keys or cloud dependencies. Use the provided tools to schedule tasks, persist memory, watch folders, and manage workers directly from your Claude prompts or Daymon UI. The built-in tools include memory storage (remember, recall, forget, memory_list), a scheduler (schedule, list_tasks, run_task, pause_task, resume_task, delete_task, task_history, task_progress, reset_session), and file watching (watch, unwatch, list_watches), plus settings controls. You can also create and manage workers (create_worker, list_workers, update_worker, delete_worker) to tailor Claude interactions to specific personalities and roles. Daymon emphasizes 100% local operation, with all data stored in a local SQLite database and no API keys or cloud services required.

Prerequisites:

• Node.js (16.x or newer) installed on your system
• Git installed to clone the repository

1. Clone the Daymon MCP server repository: git clone https://github.com/daymonio/daymon.git cd daymon

2. Install dependencies for the sidecar server: npm install

3. Build or prepare the sidecar server if required by the project (follow repository-specific build steps, typically): npm run build

4. Run the Daymon MCP sidecar server locally (as an MCP server): node path/to/sidecar/server.js

5. Verify the server is running and listening on the configured port (default 3000). You should be able to connect your MCP client to the configured endpoint.

Note: If there is a designated npm script for starting the sidecar (e.g., npm run start), you can use that instead of invoking node directly, e.g.: npm run start

6. Optional: configure environment variables (example):
   • PORT: Port to listen on (default 3000)
   • DAYMON_CONFIG: Path to a config file if you maintain custom settings

Troubleshooting tips:

• Ensure Node.js version compatibility as specified by the repository.
• Check for permission issues when watching folders or writing to SQLite files.
• If using macOS or Windows, ensure appropriate file permission and quarantine settings are addressed for downloaded components.

Tips and common issues:

• Daymon runs Claude locally; no cloud API keys are required. Ensure your Claude subscription is accessible locally as configured by Daymon.
• When using file watchers, ensure the paths are accessible by the Daymon process and that your OS permissions allow read/write.
• The system persists memory and task state in a local SQLite database. Regular backups of this file can help prevent data loss.
• If you update the MCP server, re-install dependencies and rebuild if necessary to reflect changes in the sidecar.
• Environment variables can tailor behavior (port, config paths). If you’re deploying in a container, map volumes for persistent storage and configure the port exposure accordingly.
• For advanced use, you can define multiple workers with distinct system prompts to handle different tasks (research, code review, data analysis, etc.).