deepgram-go-voice-agent

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

• Third-party content exposure detected (high risk: 0.75). The required runtime workflow is a Deepgram Voice Agent WebSocket session where the server sends event payloads (e.g., WelcomeResponse, ConversationTextResponse, FunctionCallRequestResponse) and the client forwards/handles them as LLM-relevant message text; this is outsider-authored free text originating from the Deepgram service over the network.