notebooklm

[Skill Scanner] Installation of third-party script detected The skill's stated purpose aligns with the capabilities described (browser automation, local auth storage, and notebook querying). However, the design concentrates sensitive actions around an authoritative installer (run.py) that auto-creates environments and downloads dependencies and Chromium without showing exact sources in this document. That centralization — combined with persistent storage of browser cookies and auth_info — increases supply-chain and credential-exfiltration risk if any script or dependency is malicious or compromised. I rate this as SUSPICIOUS rather than benign: the functionality is plausible and can be legitimate, but the installer-centric workflow and lack of explicit network/source transparency warrant caution and code review of run.py, requirements.txt, and the automation scripts before running. Recommend reviewing the actual run.py and all dependency URLs, verifying requirement pins, and confirming network endpoints used by the automation (ensure direct calls to Google/Gemini only) before trusting the skill with real Google accounts or private notebooks. LLM verification: The SKILL.md content is largely coherent with the claimed purpose (querying NotebookLM via browser automation and managing a local notebook library). There is no direct evidence of embedded malware in the documentation itself. However the skill requires automatic creation of a virtual environment and installation of dependencies via run.py and performs browser automation that can access Google session data and full notebook contents. Those capabilities are proportionate to the task but expand th