notebooklm

SUSPICIOUS. The skill’s capabilities generally match its stated NotebookLM purpose, but its trust model is weak: it installs a personal third-party CLI through an unseen script, uses cookie extraction instead of official OAuth/API flows, and relies on undocumented internal APIs. This is not confirmed malware, but it creates high supply-chain and credential-forwarding risk that is disproportionate compared with an official integration.