Todo

You can run a Model Context Protocol (MCP) server that connects AI assistants to the Todo for AI task management system. It lets you retrieve tasks, fetch project details, create new tasks, and submit feedback through a standardized interface, enabling real-time updates and secure access for your AI workflows.

How to use

You will interact with the MCP server from an MCP client. Start by choosing the transport you prefer. HTTP transport provides a standalone HTTP server with REST-like endpoints and Server-Sent Events (SSE) for real-time updates. Or run the local MCP client alongside your IDE if you use a stdio-based workflow. From your client, initialize a session, then perform actions such as listing project tasks, retrieving task details, creating new tasks, and submitting feedback. The server handles authentication, session management, and secure communication, so you can focus on building AI-driven task management flows.

How to install

Prerequisites: you need Node.js 18+ and a working Todo for AI backend.

# Install the MCP server globally
npm install -g @todo-for-ai/mcp

# Start the MCP server with default HTTP transport
todo-for-ai-mcp --api-token your-token

Configuration and usage notes

Configure authentication, API base URL, timeouts, and security options to fit your environment. You can provide values via environment variables or a config file, and CLI arguments take precedence over environment variables.

{
  "apiBaseUrl": "https://todo4ai.org/todo-for-ai/api/v1",
  "apiTimeout": 10000,
  "apiToken": "your-api-token",
  "logLevel": "info"
}

Security and session management

The MCP server supports DNS rebinding protection, CORS controls, and origin validation. Sessions are created when you initialize a connection and automatically expire after the configured timeout. Keep your API token secure and rotate credentials as needed.

Troubleshooting and tips

If you encounter connection issues, verify that the backend Todo for AI service is reachable and that the API base URL is correct. Check authentication tokens, and ensure the MCP server process has network access. For debugging, enable debug logging and inspect the logs for any clues.

Available tools

get_project_tasks_by_name

Fetch all pending tasks for a given project name with optional status filtering.

get_task_by_id

Retrieve detailed information for a specific task by its ID.

create_task

Create a new task with full metadata in a specified project.

submit_task_feedback

Submit feedback and update the status for a task.

get_project_info

Retrieve detailed statistics and recent tasks for a project.