Notebooklm

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. Drastically reduced hallucinations through document-only responses.

Install only if you are comfortable letting the skill automate a logged-in Google/NotebookLM browser session and store reusable session data locally. Use a dedicated Google account and non-sensitive notebooks, protect or delete the stored browser state when finished, avoid public 'anyone with link' notebooks for confidential material, and do not rely on the skill for workflows that must comply strictly with Google service limits or enterprise data-handling rules.

Type: OpenClaw Skill Name: notebooklm-skill Version: 0.1.0 The skill is classified as suspicious due to its extensive use of high-risk capabilities, even though the stated intent appears functional. Key indicators include: 1) Browser automation with anti-detection features (e.g., `StealthUtils`, `BROWSER_ARGS` in `scripts/browser_utils.py` and `scripts/config.py`), which, while explained as necessary for reliable interaction with Google, are commonly associated with malicious bots. 2) Local storage of sensitive Google authentication session tokens (`state.json`) in `~/.claude/skills/notebooklm/data/browser_state/` as confirmed by `scripts/auth_manager.py` and `AUTHENTICATION.md`. 3) Significant use of prompt injection against the AI agent in `SKILL.md` and `scripts/ask_question.py` (e.g., 'CRITICAL: Add Command - Smart Discovery', 'EXTREMELY IMPORTANT: Is that ALL you need to know?'), instructing the agent to perform specific actions and decision-making, which is a high-risk technique even if used for benign purposes here. 4) The skill installs a full browser (Google Chrome) via `patchright install chrome` as seen in `scripts/setup_environment.py` and `requirements.txt`, and uses `subprocess.run` for various system commands, granting it powerful execution capabilities.