limitless-ai

Limitless AI MCP Server exposes five MCP tools that bridge your AI assistant with Limitless Pendant recordings. It provides resources and prompts to search, list, and retrieve lifelogs, including the ability to fetch markdown content and headings for richer analysis. To use it, install and run the server, then connect your MCP client (Claude Desktop, Windsurf, Cursor, or other MCP-compatible clients). Once configured, you can call tools like limitless_list_recent_lifelogs to fetch the latest recordings, or limitless_search_lifelogs to perform keyword searches across your recordings. The resources feature allows browsing lifelogs via URIs such as lifelog://recent or lifelog://2024-01-15, enabling structured access to recordings and metadata. The server also supports automated capability exposure so clients can discover available tools automatically.

Typical workflow: start the server with a valid LIMITLESS_API_KEY, then configure your MCP client to point at the node process. Use the tool for listing, searching, and retrieving by ID, and leverage the prompt templates to guide analysis tasks (summaries, trends, or investigations across pendant data). The integration enables rapid interrogation of your Limitless Pendant data directly from your AI assistant, with caching and retry logic to improve reliability.

Prerequisites:

• Node.js v22+ installed on your system
• Git installed
• Access to a Limitless API key (LIMITLESS_API_KEY)

Step-by-step installation:

1. Clone the repository

git clone https://github.com/ericbuess/limitless-ai-mcp-server.git
cd limitless-ai-mcp-server

2. Install dependencies

npm install

3. Build the project

npm run build

4. Set up your Limitless API key and optional settings

# Add to your shell profile
export LIMITLESS_API_KEY="your-api-key-here"
export LIMITLESS_TIMEOUT=120000      # optional: timeout in ms
export LIMITLESS_BASE_URL="https://api.limitless.ai/v1"  # optional: override API endpoint
export LOG_LEVEL="INFO"            # optional: DEBUG, INFO, WARN, ERROR
export CACHE_MAX_SIZE=100
export CACHE_TTL=300000
export SEARCH_CACHE_MAX_SIZE=50
export SEARCH_CACHE_TTL=180000

5. Run the MCP server locally

# Start the server (production setups may use a process manager)
npx serve -y limitless-ai-mcp-server  # placeholder if using a runner; otherwise run the next step directly

6. Run the built server file directly (typical Node deployment)

node /path/to/limitless-ai-mcp-server/dist/index.js

7. Optional: configure Claude Desktop or Claude Code CLI per the README configuration examples (see how_to_use for usage cues).

Tips and considerations:

• Ensure LIMITLESS_API_KEY is present in the environment where the server runs. Many clients expect this to be available when initiating the MCP connection.
• The server exposes five core MCP tools: limitless_get_lifelog_by_id, limitless_list_lifelogs_by_date, limitless_list_lifelogs_by_range, limitless_list_recent_lifelogs, limitless_search_lifelogs. Each tool supports includeMarkdown and includeHeadings to control content richness.
• Caching (CACHE_MAX_SIZE, CACHE_TTL) and search caching settings help improve performance for frequent queries; tuning may be needed for large datasets.
• If you see timeouts or authentication errors, verify LIMITLESS_API_KEY and LIMITLESS_BASE_URL are correctly set, and that the Limitless Pendant data is accessible with the provided API key.
• The MCP client can auto-discover capabilities; ensure your client is configured to query the running node process and that the environment variable LIMITLESS_API_KEY is visible to the client when invoking the server.
• When upgrading the server or dependencies, re-run npm install and npm run build to ensure dist/index.js is up-to-date.