hucompany

Key Features

Related MCP server: NotebookLM MCP Server

Important Notice

This MCP uses the following internal APIs:

• They are undocumented and may change without notice.

• It requires extracting cookies from your browser.

Please use it only for personal/experimental purposes.

Installation

You can install notebooklm-mcp-server from PyPI using your preferred Python package manager (uv, pip, pipx, etc.).

You can also install directly from the source code.

Updating

Use your package manager's upgrade command to update to the latest version.

After updating, restart your AI tool to refresh the connection:

• Claude Code: Restart the application or use the reconnect command

• Cursor: Restart the application

• Gemini CLI: Restart the CLI session

Uninstallation

To completely remove the MCP, use your package manager's uninstall command. You can also optionally remove cached authentication tokens or entries from your AI tool's configuration files.

Authentication

You need to authenticate with NotebookLM before using the MCP.

Automatic mode is recommended. It launches a dedicated Chrome profile; when you sign in to Google, cookies are extracted automatically. Your login information is saved for future authentication refreshes.

If automatic mode does not work, you can manually extract and save cookies from Chrome DevTools using file mode.

After successful authentication, add the MCP to your AI tool and restart it.

For details and troubleshooting, see **docs/AUTHENTICATION.md**.

MCP Configuration

⚠️ Context Window Caution: This MCP provides 31 tools, so it can consume a large portion of the context window. When you are not actively using NotebookLM, it is recommended to disable the MCP to free up context for other tasks.

CLI Options

You can configure the server using command-line arguments. You can set the transport protocol (stdio, http, sse), port, host, debug mode, query timeout, and more.

Environment Variables

You can also configure it through environment variables. (NOTEBOOKLM_MCP_TRANSPORT, NOTEBOOKLM_MCP_PORT, etc.)

Connecting AI Tools (Claude Code, Gemini CLI, IDE, etc.)

Tools such as Claude Code and Gemini CLI can easily add the MCP server using built-in commands.

For tools that use JSON configuration files, such as Cursor, VS Code, and Claude Desktop, you need to find the path of the installed notebooklm-mcp and add it to the mcpServers entry in the configuration file.

Usage Examples

You can converse in natural language with AI tools such as Claude Code, Cursor, and Gemini CLI and use it in the following ways.

Research and Discovery

• "Show me a list of all my NotebookLM notebooks"

• "Create a new notebook called 'AI Strategy Research'"

• "Start web research on 'Enterprise AI ROI Metrics' and show me the sources found"

• "Conduct in-depth research on 'Cloud Marketplace Trends' and import the top 10 sources"

• "Search my Google Drive for documents related to 'Product Roadmap' and create a notebook"

Adding Content

• "Add this URL to my notebook: https://example.com/article"

• "Add this YouTube video about Kubernetes to my notebook"

• "Add my meeting notes to this notebook as a text source"

• "Import this Google Doc into my research notebook"

AI-Powered Analysis

• "What are the key findings in this notebook?"

• "Summarize the main claims across all these sources"

• "What does this source say about security best practices?"

• "Give me an AI summary of what this notebook is about"

• "Change the chat settings to study guide style and make the answers longer"

Content Generation

• "Turn this notebook into an audio podcast overview in deep-dive format"

• "Generate a video overview with a classic visual style"

• "Create a briefing document based on these sources"

• "Make flashcards for studying at medium difficulty"

• "Generate a landscape infographic"

• "Draw a mind map from my research sources"

• "Create a presentation slide deck from this notebook"

Smart Management

• "Check which Google Drive sources need updating and sync them"

• "Show me all the sources in this notebook and whether they are up to date"

• "Delete this source from the notebook"

• "Check the status of my audio overview generation"

Authentication Lifecycle

• Cookies: Last about 2~4 weeks (automatically renewed through headless Chrome if a saved profile exists)

• CSRF token: Lasts about a few minutes (automatically renewed when a request fails)

• Session ID: Persists per MCP session (automatically extracted when the MCP starts)

v0.1.9+: The server automatically handles token expiration.

1. Instantly renews the CSRF token when it expires

2. Reloads cookies from disk if they were updated externally

3. Runs headless Chrome authentication if a saved login profile exists

If automatic renewal fails (when your Google login has fully expired), you need to run the authentication tool again.

Troubleshooting

If Updates Are Not Applied Properly

The latest version may not be installed due to package manager cache issues or version constraints. In this case, you can resolve it by using the force reinstall option.

If Chrome DevTools MCP Is Not Working (Cursor/Gemini CLI)

If Chrome DevTools MCP appears as "no tools" or fails to start, it may be a known npx bug. In that case, changing the configuration to run it with pnpm may resolve the issue.

Limitations

• Rate limit: The free tier is limited to about 50 queries per day.

• No official support: The API may change without notice.

• Cookie expiration: You may need to re-extract cookies every few weeks.

Contributing

See CLAUDE.md for API documentation and how to add new features.

License

MIT License