notebooklm

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). This skill automates a browser to query Google NotebookLM at notebooklm.google.com and ingests user-uploaded notebooks (added via notebook URLs) as part of its workflow, so untrusted user-generated content from those notebooks could be read and influence agent answers/actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly uses browser automation at runtime to access NotebookLM (e.g., https://notebooklm.google.com/notebook/YOUR_ID) and injects notebook content into the agent's context to produce answers, meaning remote content from that URL directly controls the agent's prompts/output.