deepgram-rust-voice-agent
Installation
SKILL.md
Using Deepgram Voice Agent (Rust SDK)
Use this skill when the user wants an end-to-end conversational voice agent — STT, LLM, and TTS orchestrated by Deepgram over one WebSocket.
When to use this product
- Building a full conversational agent (the server runs the turn loop, the LLM call, and speech synthesis).
- Handling agent events: conversation text, function-call requests, latency reports, interruptions.
- Changing a live session's model, prompt, or voice without reconnecting.
Prefer conversational STT (flux_request) instead when the application only needs turn-based transcripts and drives its own LLM and TTS.
Authentication
The Voice Agent client is behind the agent feature (on by default).