longform-video-clone-edit

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly downloads arbitrary YouTube videos (Step 1: yt-dlp "YOUTUBE_URL") and then transcribes and analyzes those user-generated/public video frames and text (Steps 2, 6–8) to drive segmentation, HeyGen generation, and compositing decisions, so untrusted third‑party content can directly influence agent actions.