notebooklm

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly ingests public third-party content (e.g., via source_add(type="url") and YouTube, research_start -> research_import for web discovery described in rules/workflow-research-discovery.md and the Decision Tree), and that untrusted content is read and used by notebook_query and studio pipelines to drive subsequent actions, so external pages/posts can materially influence agent behavior.