media

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

• Third-party content exposure detected (medium risk: 0.65). Outsider free text can be ingested into the agent’s LLM context via NotebookLM when the user supplies an outsider URL/source (e.g., generate.py → client.sources.add_url(...) / notebooklm source add ...), causing NotebookLM to fetch and index web content that may include prompt-injection text.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly runs notebooklm source add "https://..." (for example https://arxiv.org/abs/2401.12345 and https://www.youtube.com/watch?v=VIDEO_ID) and notebooklm source add-research at runtime, which fetches and indexes remote content that is then used by NotebookLM to drive generation—i.e., external URLs fetched at runtime directly influence the agent's prompts/outputs.