livekit-voice

What This Skill Does

The livekit-voice skill provides a robust command-line interface for managing real-time audio and video infrastructure within OpenClaw. It allows you to programmatically create and tear down rooms, manage participant access through JWT generation, monitor live room status, and trigger server-side recordings (Egress). Whether you are building a voice-enabled AI assistant, setting up a real-time conferencing bridge, or streaming audio from TTS services like ElevenLabs, this skill serves as the control plane for your WebRTC infrastructure.

Installation

To install this skill, use the following command in your terminal: clawhub install openclaw/skills/skills/aiwithabidi/livekit-voice

Ensure you have your LiveKit credentials ready. You will need to export LIVEKIT_API_KEY, LIVEKIT_API_SECRET, and LIVEKIT_URL into your environment variables to allow the agent to authenticate with your LiveKit server or cloud instance.

Use Cases

• Voice AI Agents: Orchestrate real-time communication for agents using OpenAI Realtime or Vapi by creating ephemeral rooms on demand.
• Event Recording: Automatically trigger room recordings whenever a support call or meeting begins to archive interactions for training or compliance.
• Access Control: Generate secure, time-limited JWT access tokens for clients or participants joining your application, ensuring only authorized users can publish or subscribe to streams.
• Automated Infrastructure: Use as part of a larger CI/CD or automation workflow to clean up idle rooms and manage resource usage on your LiveKit server.

Example Prompts

1. "Create a new LiveKit room named 'support-session-001' with a maximum of 5 participants and set the auto-close timeout to 600 seconds."
2. "Generate an access token for a user named 'Alice' with ID 'user_88' for the room 'training-room', allowing them to publish audio and subscribe to existing tracks."
3. "List all active rooms currently running on our server and then start an egress recording for 'main-conference-room' saving to our s3://logs-bucket/ recordings folder."

Tips & Limitations

• Security: Always rotate your LIVEKIT_API_SECRET periodically and store it in a secret management service rather than hard-coding it into scripts.
• Egress: Note that recording requires an active Egress service configured on your LiveKit server; ensure this is enabled in your server configuration, otherwise the record command will fail.
• Latency: Because this skill interacts with a remote WebRTC server, ensure your LIVEKIT_URL is reachable from your OpenClaw environment. If you are self-hosting, ensure ports 7880 and 7881 are accessible.