youtube-channel-breakdown

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

• Third-party content exposure detected (high risk: 1.00). The skill's Step 1 runs a Python research pipeline that fetches all videos/comments from a public YouTube channel (via channel_research_pipeline.py) and Step 2 explicitly instructs the agent to read untrusted output files like research_data/[ChannelName]_/top_comments.json, top_videos.json, and analysis_summary.json — public, user-generated content that the agent must interpret to drive its analysis and actions, creating clear exposure to indirect prompt injection.