notebooklm

Query Google NotebookLM notebooks via browser automation for source-grounded answers.

Updated Apr 7, 2026
One-click install
npx skills add https://github.com/NT-boop-star/BRMV-tract --skill notebooklm-nt-boop-star
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: notebooklm
Source: https://github.com/NT-boop-star/BRMV-tract/tree/main/antigravity/skills/notebooklm
Command: npx skills add https://github.com/NT-boop-star/BRMV-tract --skill notebooklm-nt-boop-star

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires patchright, python-dotenv, and includes scripts (resource) and references (resource) components.

What problem does it solve? Searching local documentation with an AI assistant consumes large amounts of tokens, misses context across files, and often produces hallucinated APIs. This Skill lets Claude Code ask questions directly to Google NotebookLM, which answers exclusively from your uploaded documents with citation-backed responses from Gemini. ## Core Features & Use Cases - Direct NotebookLM Queries: Ask questions to any shared NotebookLM notebook through automated Chrome browser sessions using Patchright, with human-like typing and anti-detection behavior. - Notebook Library Management: Save, search, tag, and activate notebooks with metadata so the right knowledge base is selected automatically for each task. - Persistent Google Authentication: One-time manual Google login with a hybrid browser-profile plus cookie-injection approach that keeps sessions valid across restarts. - Use Case: While building an n8n workflow, ask your n8n documentation notebook about node configurations and receive accurate, document-grounded answers instead of hallucinated API calls. ## Quick Start Ask Claude to set up NotebookLM authentication, then say "Ask my React docs notebook what the rules of hooks are" to get a source-grounded answer.

Frequently Asked Questions about notebooklm

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I query NotebookLM from Claude Code?

Run the ask_question.py script through the run.py wrapper with your question and a notebook ID or URL. The skill opens a browser session, submits the question to NotebookLM, waits for the response to stabilize, and returns the answer text.

How does NotebookLM authentication work with browser automation?

Authentication uses a one-time manual Google login in a visible Chrome window. The skill saves both a persistent browser profile and a state.json cookie file, then injects cookies manually at runtime to work around a Playwright session-cookie persistence bug.

Does this skill work in the Claude web UI?

No, it only works with local Claude Code installations. The web UI runs skills in a sandbox without network access, which browser automation requires to reach NotebookLM.

What is the difference between this skill and the NotebookLM MCP server?

The skill is a Python implementation running directly in Claude Code with stateless, fresh browser sessions per question. The MCP server is TypeScript-based, supports persistent chat sessions, and works with multiple tools like Codex and Cursor.

What are the rate limits for querying NotebookLM?

Free Google accounts are limited to roughly 50 queries per day. When you hit the limit, you can wait for the daily reset or re-authenticate with a different Google account using the reauth command.

Why does the skill fail with ModuleNotFoundError?

Scripts must be executed through the run.py wrapper, which creates the virtual environment and installs dependencies automatically. Calling scripts directly with python scripts/script_name.py fails because the isolated .venv is not active.