Node.js Multimodal Voice Agent

Deploy a sandbox app • LiveKit Agents Docs • LiveKit Cloud • Blog

A basic example of a multimodal voice agent using LiveKit and the Node.js Agents Framework.

Dev Setup

Clone the repository and install dependencies:

pnpm install

Set up the environment by copying .env.example to .env.local and filling in the required values:

• LIVEKIT_URL
• LIVEKIT_API_KEY
• LIVEKIT_API_SECRET
• OPENAI_API_KEY

You can also do this automatically using the LiveKit CLI:

lk app env

To run the agent, first build the TypeScript project, then execute the output with the dev or start commands:

pnpm build
node dist/agent.js dev # see agents-js for more info on subcommands

This agent requires a frontend application to communicate with. You can use one of our example frontends in livekit-examples, create your own following one of our client quickstarts, or test instantly against one of our hosted Sandbox frontends.