notebooklm

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly opens and scrapes NotebookLM pages via browser automation (see SKILL.md "Ask Questions" flow and scripts/ask_question.py which navigates to a provided --notebook-url and extracts the answer), and those NotebookLM notebooks can be public or contain user-uploaded/linked web content that the agent reads and uses to drive follow-up queries and decisions, so untrusted third‑party content can materially influence the agent's actions.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill directly opens NotebookLM notebook URLs at runtime (e.g., https://notebooklm.google.com/ and https://notebooklm.google.com/notebook/...) via page.goto(...) and consumes the NotebookLM responses to drive and synthesize agent outputs, so external content directly controls the agent's prompts/answers.