groq-media-transcript

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow scripts/transcribe.sh (when given a public page URL) uses yt-dlp to resolve/play media from attacker-controlled webpages/metadata and then processes extracted stream/audio output, which can include outsider-authored text and causes the LLM to ingest it via Groq transcription results returned from that source.