A Model Context Protocol server for accessing Limitless AI lifelog data
npm install mcp-limitless

A Model Context Protocol (MCP) server that connects AI assistants to Limitless AI lifelog data. This server enables seamless integration with your recorded conversations, thoughts, and daily activities from your Limitless AI pendant, allowing AI assistants to read, search, and analyze your personal lifelog.
Limitless AI creates a continuous lifelog by recording your conversations and activities throughout the day. This MCP server bridges the gap between your lifelog data and AI assistants, enabling:
- Personal Memory Access: Retrieve specific conversations and moments from your lifelog
- Intelligent Search: Find relevant content across your entire recorded history
- Timeline Analysis: Understand patterns and evolution in your daily activities
- Context-Aware Assistance: AI assistants can reference your past conversations and experiences
- Limitless AI Integration: Direct access to your lifelog data via the official API
- Type-Safe Implementation: Written in TypeScript with comprehensive type definitions
- Input Validation: Robust validation for all API inputs using Zod schemas
- Error Handling: Graceful error handling with informative messages
- MCP Integration: Full compatibility with Claude, Cursor, Windsurf, Cline, and other MCP hosts
- getLifelogs: Retrieve lifelog entries with flexible filtering options:
- Filter by date, timezone, and time ranges
- Pagination support for large datasets
- Sort by ascending or descending order
- getLifelogEntry: Get detailed information about a specific lifelog entry by ID
- searchLifelogs: Search through your lifelog content:
- Full-text search across summaries and content
- Date range filtering
- Pagination for comprehensive results
The easiest way to install Limitless AI MCP is using Smithery:
``bashFor Claude Desktop
npx -y @smithery/cli install @Hint-Services/mcp-limitless --client claude
$3
`bash
npm install mcp-limitless
`Configuration
Add the server to your MCP settings file with the following configuration:
`json
{
"mcpServers": {
"limitless": {
"command": "npx",
"args": ["-y", "mcp-limitless"],
"env": {
"LIMITLESS_API_KEY": "your-limitless-api-key"
}
}
}
}
`$3
-
LIMITLESS_API_KEY: Your Limitless AI API key (get it here)$3
-
LIMITLESS_BASE_URL: Custom API base URL (defaults to https://api.limitless.ai)$3
1. Visit Limitless AI Developer Platform
2. Sign in to your Limitless account
3. Navigate to Developer settings
4. Generate a new API key
5. Important: Never share your API key or commit it to source control
Example Workflows
$3
`json
{
"tool": "getLifelogs",
"arguments": {
"date": "2024-01-15",
"timezone": "America/Los_Angeles",
"limit": 10
}
}
`$3
`json
{
"tool": "searchLifelogs",
"arguments": {
"query": "meeting with Sarah about project planning",
"limit": 5
}
}
`$3
`json
{
"tool": "getLifelogs",
"arguments": {
"date": "2024-01-15",
"start_time": "09:00:00",
"end_time": "17:00:00",
"timezone": "America/New_York"
}
}
`$3
`json
{
"tool": "getLifelogEntry",
"arguments": {
"lifelog_id": "entry_12345"
}
}
`$3
`json
{
"tool": "searchLifelogs",
"arguments": {
"query": "quarterly planning",
"date_from": "2024-01-01",
"date_to": "2024-01-31",
"limit": 10
}
}
`API Limitations
- The Limitless AI API is currently in beta
- Maximum 10 entries per request
- Only supports Pendant device data currently
- Search functionality is implemented locally (API search coming soon)
Project Structure
`
mcp-limitless/
├── src/
│ ├── index.ts # Main MCP server entry point
│ └── limitless/ # Limitless AI integration
│ ├── client.ts # Limitless client implementation
│ └── types.ts # TypeScript type definitions
├── docs/ # Documentation
├── package.json # Project configuration
└── tsconfig.json # TypeScript configuration
`For Developers
$3
-
pnpm install - Install dependencies
- pnpm run build - Build the project for HTTP streaming (default)
- pnpm run dev - Run in development mode with hot reloading
- pnpm run inspector - Launch MCP inspector for testing (stdio mode)
- pnpm run test - Run tests$3
#### HTTP Streaming (Recommended)
`bash
pnpm run build:http
pnpm run start:http
`#### Traditional stdio (Legacy)
`bash
pnpm run build:stdio
pnpm run start:stdio
`The HTTP streaming interface offers better performance and simplified deployment without requiring Docker, while maintaining full backwards compatibility with the traditional stdio interface.
$3
The server implements the following MCP tools:
-
getLifelogs - Retrieve lifelog entries with filtering
- getLifelogEntry - Get specific lifelog entry by ID
- searchLifelogs - Search through lifelog contentFor detailed API documentation, see the Limitless AI Developer Documentation.
Privacy & Security
- Your API key is used only to authenticate with Limitless AI's official API
- No data is stored or cached by this MCP server
- All communication is directly between your client and Limitless AI's servers
- Follow Limitless AI's privacy policy and terms of service
Troubleshooting
$3
- Ensure your API key is valid and not expired
- Check that you're using the correct environment variable name: LIMITLESS_API_KEY`For further information, refer to:
- Limitless AI Developer Documentation: Official API documentation
- Model Context Protocol Documentation: MCP architecture and design principles
- Smithery - MCP Server Registry: Guidelines for publishing MCP servers
- MCP TypeScript SDK Documentation: Comprehensive TypeScript SDK documentation
> "The future is already here, it's just unevenly distributed"
>
> — William Gibson, Author
Hint Services is a boutique consultancy with a mission to develop and expand how user interfaces leverage artificial intelligence technology. We architect ambition at the intersection of AI and User Experience, founded and led by Ben Hofferber.
We offer specialized AI workshops for design teams looking to embrace AI tools without becoming developers. Learn more about our training and workshops.