notebooklm

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/ask_question.py, the agent ingests and reads free-text from the browser-rendered NotebookLM page corresponding to user-supplied --notebook-url/library notebook content (outsider-authored text can be embedded in those notebooks), via page.query_selector_all(...).inner_text() when extracting the answer.