video-scene-analysis

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

• Third-party content exposure detected (high risk: 0.85). The required runtime workflow ingests outsider-authored free text via local video audio transcription: scripts/analyze_video.py extracts audio with ffmpeg and runs transcribe_local() using faster_whisper, then includes the resulting transcript text in the generated .analysis.md/JSON (agent later reads/uses it), so the LLM context can contain arbitrary spoken text from the input video.