notebooklm

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.70). Most links are to the official NotebookLM site and related GitHub code (low risk), but the presence of a strangely-named GitHub repo referencing anti-detection/patchright tooling (Kaliiiiiiiiii-Vinyzu), a proxy placeholder, and references to installing browser components/third‑party forks raises red flags for potential misuse or supply-chain risk, so treat this set as moderately suspicious.

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

• Third-party content exposure detected (high risk: 0.85). At runtime, scripts/ask_question.py uses Playwright to open a user-supplied --notebook-url (public web content) and then reads the resulting NotebookLM page text (including any outsider-authored free text from that notebook) into the agent context via inner_text() from RESPONSE_SELECTORS.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill opens and scrapes user-provided NotebookLM URLs at runtime (e.g., https://notebooklm.google.com/ and https://notebooklm.google.com/notebook/...) to retrieve NotebookLM answers that directly determine the agent's responses, so this external content is fetched during runtime and directly controls the agent output.