hebrew-podcast-postproduction

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

• Third-party content exposure detected (medium risk: 0.30). The workflow reads outsider-authored free text at runtime from user-supplied Whisper/ElevenLabs Scribe JSON and SRT cue text (e.g., scripts/score_segments.py loads transcript JSON and scripts/generate_bundle.py emits SRT/show notes/captions from provided cue text), so an attacker who can supply those inputs can inject arbitrary text into the agent’s processing.