Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jacob-bd/notebooklm-mcp-cli/llms.txt

Use this file to discover all available pages before exploring further.

The NotebookLM MCP server is a standards-compliant Model Context Protocol server that gives any MCP-compatible AI assistant — Claude Code, Gemini CLI, Cursor, Windsurf, GitHub Copilot, and others — direct, programmatic access to Google NotebookLM. Once connected, your AI can create and manage notebooks, add sources from the web or Google Drive, run deep research, generate studio content (podcasts, videos, slide decks, and more), share notebooks, and handle dozens of other tasks — all through natural language, without you ever opening a browser.
Context window notice: The MCP server registers 39 tools with your AI assistant. Each tool consumes a small amount of context on every request. Disable the server when you are not actively using NotebookLM to preserve context budget. In Claude Code, type @notebooklm-mcp to toggle it on or off.
Server name matters: Always register the server as notebooklm-mcp (not notebooklm). If you have a legacy browser-automation NotebookLM MCP installed under a different name, remove it first — AI assistants get confused when two servers expose overlapping tool names such as notebook_create or source_add.

Quick Setup

1

Install the package

Install notebooklm-mcp-cli from PyPI. The package includes both the nlm CLI and the notebooklm-mcp server binary.
uv tool install notebooklm-mcp-cli
You can also use pip install notebooklm-mcp-cli or pipx install notebooklm-mcp-cli.
2

Authenticate with NotebookLM

Run the login command. It launches a dedicated browser profile, you sign in with your Google account, and cookies are extracted automatically.
nlm login
Your session persists and auto-refreshes. Run nlm login --check at any time to verify your auth status.
3

Configure your AI tool

Use the nlm setup add command for automatic, zero-edit configuration — no JSON files to touch. Pick the tab for your tool:
Automatic (recommended):
nlm setup add claude-code
Or use the Claude CLI directly:
claude mcp add --scope user notebooklm-mcp notebooklm-mcp

Supported AI Tools

Claude Code

Full MCP support. Use @notebooklm-mcp to toggle. Register with nlm setup add claude-code or claude mcp add --scope user notebooklm-mcp notebooklm-mcp.

Gemini CLI

Full MCP support. Register with nlm setup add gemini or gemini mcp add --scope user notebooklm-mcp notebooklm-mcp.

Cursor

Full MCP support via ~/.cursor/mcp.json. Register with nlm setup add cursor.

Windsurf

Full MCP support via ~/.codeium/windsurf/mcp_config.json. Register with nlm setup add windsurf.

GitHub Copilot (VS Code)

Workspace-scoped MCP support via .vscode/mcp.json. Register with nlm setup add github-copilot.

Any MCP-compatible tool

Use nlm setup add json to generate a config snippet, or point the tool at the notebooklm-mcp binary directly. HTTP and SSE transports are also available.

Natural Language Examples

Once connected, you interact with NotebookLM entirely through natural language. Here are some prompts you can say to your AI assistant:
  • “List all my NotebookLM notebooks and show me their source counts.”
  • “Create a new notebook called ‘Competitive Analysis Q3’ and add this URL as a source: https://example.com/report
  • “Do a deep web research on ‘enterprise AI ROI metrics’, wait for it to finish, and import only the cited sources.”
  • “Generate an audio podcast in deep-dive format for the notebook I just created, then poll until it’s done and download the MP3.”
  • “Query the ‘Product Strategy’ notebook: what are the three most important takeaways?”
  • “Make the ‘Public Research’ notebook publicly accessible and give me the share link.”
  • “Tag all my research notebooks with ‘ai,research’, then batch-generate briefing docs for all of them.”
  • “Run the ingest-and-podcast pipeline on the ‘Weekly Reads’ notebook using https://example.com/article as input.”
Queries sent through the MCP automatically persist in your NotebookLM web UI chat history, so you can pick up the conversation in the browser at any time.

Tool Categories

The server exposes 39 tools organized into the following categories. See the Tools Reference for full parameter documentation and examples.
CategoryToolsWhat it covers
Notebooks6Create, list, get, describe, rename, delete
Sources7Add (URL/text/file/Drive), list Drive, sync Drive, rename, delete, describe, get content
Querying4Ask AI about notebook sources, async query start/status, configure chat behavior
Studio4Create audio/video/reports/quizzes/flashcards/slides/infographics, check status, delete, revise
Downloads1Download any artifact type to a local file
Exports1Export artifacts to Google Docs or Sheets
Research3Start web/Drive research, poll progress, import discovered sources
Notes1Create, list, update, and delete notebook notes (unified tool)
Labels1Auto-label and organize sources into thematic categories (unified tool)
Sharing4Get sharing status, enable/disable public link, invite collaborator, batch invite
Auth2Reload tokens from disk, save cookies manually (fallback)
Server1Version info and auth status
Batch & Cross-Notebook2Batch operations across multiple notebooks, aggregated cross-notebook queries
Pipelines1List and run multi-step automated workflows
Tags1Tag notebooks and smart-select by tag match

Build docs developers (and LLMs) love