モデルコンテキストプロトコル (MCP)

Model Context Protocol (MCP) overview

Model Context Protocol (MCP) is an open standard that connects external AI assistants like Claude Desktop, Cursor, and Cline directly to your personal u0life workspace. Through MCP, your AI assistant can inspect planned tasks, browse projects, manage clients, and create notes in real time without manual copy-pasting.

Unlike traditional generic API integrations, u0life MCP integration uses dedicated AI Bot tokens generated in the Connects section. Every action performed by your AI assistant is automatically attributed to your chosen bot, giving you crystal-clear visibility into what you did versus what your AI assistant did.

Direct AI Workspace Access Let Claude, Cursor, or Cline inspect your planned tasks, search notes, and check project context without manual copying.
Bot Action Attribution All tasks created or advanced by the AI are attributed to your bot in the activity feed and task history with its avatar and badge.
Secure & Private Runs via local stdio transport with user-owned bot tokens. Planned and in-work schedule dates are strictly protected.
Zero Extra Dependencies Lightweight Node.js stdio server runs out of the box with standard JSON-RPC 2.0 without heavy external packages.

Where to find and generate your bot token

Tokens for MCP integrations are managed within the Connects section of your u0life workspace:

  1. Step 1: Open Connects. Navigate to the Connects section in the main app menu. Go to Connects
  2. Step 2: Select or create an AI Bot. Open the profile or settings of any existing bot, or add a new custom AI bot.
  3. Step 3: Locate the MCP Integration block. In the bot settings modal, scroll down to the "AI / MCP Integration (API Token)" panel.
  4. Step 4: Generate Token. Click "Generate Token" to create a fresh secure token (starts with u0_live_).
  5. Step 5: Copy Configuration. Click "Copy MCP Config" to instantly copy the ready-to-paste JSON block for your client, or copy the raw token.

Configuration for Claude Desktop

Add the u0life MCP server to your Claude Desktop configuration file (macOS: ~/Library/Application Support/Claude/claude_desktop_config.json, Windows: %APPDATA%\Claude\claude_desktop_config.json):

{
  "mcpServers": {
    "u0life": {
      "command": "node",
      "args": ["/absolute/path/to/u0life/mcp/index.js"],
      "env": {
        "U0LIFE_API_KEY": "u0_live_your_bot_token_here",
        "U0LIFE_API_URL": "https://u0life.com/v1"
      }
    }
  }
}

Configuration for Cursor IDE

In Cursor, open Settings -> Features -> MCP (or create .cursor/mcp.json in your workspace root) and add the server configuration:

{
  "mcpServers": {
    "u0life": {
      "command": "node",
      "args": ["/absolute/path/to/u0life/mcp/index.js"],
      "env": {
        "U0LIFE_API_KEY": "u0_live_your_bot_token_here"
      }
    }
  }
}

Available MCP Tools

The u0life MCP server exposes a rich suite of native tools that allow your AI assistant to query and manage your workspace:

list_tasks tools/call List tasks with optional filters: date (YYYY-MM-DD), date_from, date_to, status, and search query.
get_task tools/call Retrieve detailed information about a task by its numeric ID.
create_task tools/call Create a new task. Automatically assigns the task to the configured bot.
update_task tools/call Update task status, title, or description. Status transitions are recorded with bot attribution.
delete_task tools/call Safely delete a task by ID.
list_projects tools/call List active projects with client associations and color indicators.
list_clients tools/call List clients and contacts associated with your account.
list_notes tools/call List and search notes across the workspace.
create_note tools/call Create a documentation note or meeting summary.

Security and Scheduling Protection

In accordance with platform scheduling invariants, planned tasks (status 1) and in-work tasks (status 2) with scheduled dates are never shifted or rescheduled. You can revoke any bot token at any time in Connects with one click.