video-editing

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

• Third-party content exposure detected (medium risk: 0.30). In Phase 2 (AI STRUCTURE), the workflow reads transcript.txt generated from a provided video URL via yt-dlp (references/video-transcript.md → references/phase-commands.md), so outsider-authored transcript text from the video becomes LLM input for EDL/cuts generation.