speech-to-text

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

• Third-party content exposure detected (high risk: 0.70). The SKILL.md explicitly shows fetching audio/video from arbitrary public URLs (e.g., "audio_url": "https://audio.mp3", "video_url": "https://video.mp4"), so the agent ingests untrusted third‑party user content as part of its transcription workflow which could contain instructions that influence downstream processing (e.g., passed into caption-videos).