video-reader

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly ingests and processes user-provided video/audio (see "Use when a user sends a video/audio" and the ffmpeg/whisper extraction/transcription steps) and the agent is expected to read the extracted frames and transcripts, exposing it to arbitrary untrusted third‑party content that could carry indirect prompt injection.