notebooklm

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

• Third-party content exposure detected (high risk: 0.85). At runtime, ask_question.py navigates to a user-supplied NotebookLM URL (page.goto(validated_url)), then reads the resulting page text from NotebookLM’s UI (latest.inner_text() from RESPONSE_SELECTORS) and appends it to the agent’s LLM context; this is outsider-authored free text originating from the NotebookLM page content (not authored by the operating user).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's runtime (ask_question.py) directly navigates to and scrapes user-supplied NotebookLM pages (e.g. https://notebooklm.google.com/notebook/...), using the live page content as the answer injected into the agent's output, so external NotebookLM content fetched at runtime can directly control prompts/answers.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)