Use this skill to query your Google NotebookLM notebooks directly from Claude Code for source-grounded, citation-backed answers from Gemini. Browser automation, library management, persistent auth. Zero hallucinations, just your knowledge base.
Detected risks:
The NotebookLM skill enables users to query their Google NotebookLM notebooks directly within the Claude Code environment. This integration allows for source-grounded, citation-backed answers by leveraging Gemini's capabilities. The primary function of this skill is to automate interactions with NotebookLM, providing a seamless way to retrieve and manage notebook data through automated browser sessions. Each session queries the notebook for answers, ensuring that no hallucinations are involved—answers are based entirely on the user's knowledge base in NotebookLM.
Key features of the NotebookLM skill include easy notebook management, such as adding, searching, and activating notebooks, as well as the ability to query notebooks by specific topics. The skill supports critical functionality like authentication setup (which includes manual Google login), browser automation for querying notebooks, and library management for keeping track of multiple notebooks. The skill's robust and reliable architecture eliminates errors, such as missing dependencies, by using a `run.py` wrapper that ensures proper setup before execution.
This skill is particularly useful for professionals who regularly interact with large documentation repositories stored in NotebookLM. Researchers, content creators, and anyone who needs access to curated, organized knowledge bases will benefit from the ability to query their notebooks efficiently. By using clear prompts and following a structured process, users can leverage their notebooks as a reliable, powerful source of information to enhance their workflows.
To authenticate, use the `python scripts/run.py auth_manager.py setup` command. A browser window will open for you to manually log in to Google. Authentication is required only once for setup.
No, you must provide a description and topics for the notebook before adding it. If unsure about the content, ask the user for details on what the notebook contains and what topics it covers.
If you call scripts directly without using the `run.py` wrapper, the execution will fail, as dependencies may not be installed, and the virtual environment may not be activated. Always use the `python scripts/run.py` command to ensure proper execution.
Yes, you can query a specific notebook by passing the `--notebook-id` or `--notebook-url` parameters when using the `ask_question.py` script.
If you encounter an issue, ensure that your authentication is valid and check that the notebook ID or URL is correct. You can also use the `--show-browser` flag for debugging to see what's happening in the browser session.
Quick Setup:
.claude/skills/Repository
delorenj/skills