LiveKit Agent Skills
Official skills for building voice AI agents — architecture, testing, debugging and deployment.
About
LiveKit maintains official agent skills in the livekit/agent-skills repo, recommended in the LiveKit Agents README itself. The pack teaches coding agents the full voice-agent lifecycle: building agents with the modular pipeline (VAD → STT → LLM → TTS), handoffs and workflows, turn-level regression testing, simulation scenarios, debugging live sessions, and operating workers in production. Install with npx skills add livekit/agent-skills --skill livekit-agents, and pair it with the LiveKit Docs MCP server — the skill teaches approach, the MCP supplies current API details.
The honest picture: it's deep on LiveKit's own Agents framework and thinner on everything else — if you're building on Pipecat or raw WebRTC, look elsewhere. Voice agents also have real infrastructure costs (STT/TTS/LLM per minute), which no skill can optimize away, and the debugging skills assume you're running LiveKit's stack end to end.
What it can do
- Voice agent architecture and workflow design
- Turn-level regression testing patterns
- Simulation scenarios and outcome analysis
- Debugging live agent sessions
- Deploying and operating agent workers
Setup
Install with: npx skills add livekit/agent-skills --skill livekit-agents. Pair with the LiveKit Docs MCP server for current API details.
Prerequisites
A LiveKit project (livekit-agents SDK) and API credentials. Python or Node.js depending on your agent worker.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 45; Trending score 82/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.