
livekit / agent-skills
★ 58A skill package that teaches your agent 1 capability — every one documented and browsable below, no GitHub required · by livekit.
Each skill below is one capability this package teaches your agent. Install the whole package, or open a skill to install just that one.
LiveKit Agent Skills
Reusable skills for AI coding agents building with LiveKit. These skills provide procedural knowledge that helps agents build voice AI applications more effectively.
Available Skills
| Skill | Description |
|---|---|
| livekit-agents | Architectural guidance for building low-latency voice AI agents with LiveKit Agents SDK. Covers workflow design, handoffs, tasks, and mandatory testing practices. |
| livekit-simulations | Generate targeted test scenarios for a LiveKit agent — locally, from the agent's own code plus what the user wants stress-tested — and run them as simulations. Bundles persona/goal/challenge libraries and a stdlib scenario builder. |
Installation
npx skills add livekit/agent-skillsSkills activate automatically when agents detect relevant tasks (e.g., "build a voice agent", "create a LiveKit agent").
Skill Structure
Each skill contains:
SKILL.md— Agent instructions with behavioral guidancereferences/— Supporting documentation
Using with MCP
These skills are designed to work alongside the LiveKit Docs MCP server. The skills provide behavioral guidance (how to think about building voice agents) while MCP provides factual information (current API signatures, configuration options).
Install the MCP server for best results. Installation instructions for all supported coding agents are available at:
https://docs.livekit.io/intro/mcp-server/
Design Philosophy
These skills follow the "freeze forever" principle—they encode behavioral guidance that remains valid regardless of API changes:
- Behavior over knowledge: Skills teach how to approach problems, not API specifics
- MCP for facts: All factual information (API signatures, config options) comes from live documentation
- Testing required: Every agent implementation must include tests
- Latency-first: Voice AI requires different architectural thinking than text-based agents
Contributing
Contributions welcome! When adding or modifying skills:
- Ensure content passes the "freeze forever" test—would it still be correct if never updated?
- Direct agents to MCP/docs for all API specifics
- Include trigger phrases in the skill description
- Keep skills under 500 lines for context efficiency
License
MIT
Install the whole package (1 skills):
npx skills add https://github.com/livekit/agent-skillsOr install a single skill:
npx skills add https://github.com/livekit/agent-skills --skill <name>Pick the skill name from the Skills tab — each entry there installs independently.