elevenlabs

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

• Third-party content exposure detected (high risk: 0.90). The skill's Conversational AI workflow explicitly ingests caller audio from a websocket (see SKILL.md "async for audio_chunk in websocket: response = await agent.process_audio(audio_chunk)"), meaning it consumes untrusted user-generated audio that the agent interprets and uses to decide/respond, allowing indirect prompt injection via third-party input.