voice-ai-development

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

• Third-party content exposure detected (medium risk: 0.30). Outsider-authored free text is ingested at runtime via the WebSocket/API endpoints where user-supplied content (transcribed speech text and/or text request fields) is passed into the LLM/TTS nodes (e.g., LangGraph llm_response_node uses transcript from live transcription; Gemini ADK uses audio_bytes_hex:... derived from client-provided audio and then the ResponseAgent generates response_text from the resulting transcript).