How can I connect my local deployed agent to agents playground

Thanks for reaching out, Aditya. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

I am not using cloud livekit. I am locally runnign the livekit

Thanks — good to know you're using a self‑hosted LiveKit. I’ll walk you through the minimal steps to connect your local LiveKit + agent to the hosted Agents Playground, and then ask two quick questions to tailor troubleshooting. What to do (summary) 1. Expose your local LiveKit server publicly ◦ Start LiveKit so it binds to 0.0.0.0 (not 127.0.0.1): livekit-server --dev --bind 0.0.0.0 • Use a tunnel (eg. ngrok) to get a public URL that the Playground can reach: ngrok http 7880 • Note the public wss:// host from ngrok (e.g., wss://abc123.ngrok.io) 1. Configure your agent to use the public LiveKit endpoint ◦ Start your agent in dev mode but point its LiveKit URL to the ngrok/public URL (not localhost). ◦ Ensure it uses the same dev `api_key`/`secret` for your local server. 1. Connect the hosted Agents Playground ◦ Open https://agents-playground.livekit.io/ ◦ Enter the public LiveKit URL from the tunnel, a room name, and a valid token (or dev credentials). ◦ Create/connect the room — your agent (running in dev mode and pointed at that server) should join automatically. Quick troubleshooting checklist • Confirm LiveKit is reachable from the internet (paste the ngrok URL in a browser or use wscat). • Confirm agent logs show successful registration/connection and no auth errors. • Ensure firewall/NAT allows incoming traffic for the tunnel. • Use the same LiveKit URL, API key/secret, and room name in both agent and Playground. Two quick questions to help further • Which agent runtime are you using: Python or Node? • Do you already have a public tunnel URL (ngrok or similar) you can share? Tell me those and any agent/server log lines showing errors and I’ll give targeted next steps. Sources: Running LiveKit locally | LiveKit docs | Agents playground | LiveKit docs

Can't I run the playground also locally?

Yes — you can run the Agents Playground locally so it talks directly to your self‑hosted LiveKit and agent. Quick steps 1. Clone the repo: ◦ https://github.com/livekit/agents-playground 1. Install and start: ◦ Follow the repo README to install dependencies (npm/pnpm/yarn) and run the dev server (usually

pnpm dev

or

npm run dev

). 1. Configure the Playground to point to your LiveKit server: ◦ Update the playground's configuration or environment (per the repo README) to use your local LiveKit URL (eg.

<ws://your-host:7880>

or the public tunnel URL if needed) and a valid token or dev credentials. 1. Run your agent and LiveKit server locally: ◦ Start LiveKit binding to 0.0.0.0 if you need external access. ◦ Start your agent in

dev

mode and ensure it connects to the same LiveKit URL. Troubleshooting tips • If browser media (mic/camera) is blocked, allow permissions for the local dev origin (http://localhost:PORT). • If the playground can’t reach your LiveKit server, either run the playground and server on the same machine/network or expose LiveKit with a tunnel (ngrok) and use that public URL in the playground. • Check console and agent logs for registration/auth errors; confirm API key/secret match. Would you like step‑by‑step commands for cloning + running the playground on your OS (macOS/Linux/Windows), or help updating the playground config to your LiveKit URL? Which runtime are you using for your agent (Node or Python)? Sources: Agents playground | LiveKit docs | An open source stack for real-time multimodal AI | Voice AI quickstart | LiveKit docs

how can I make my two agents talk to each other?

You're asking how to make two agents talk to each other — I can help. Short answer: allow agents to accept audio from other agents and ensure both are in the same LiveKit room with TTS enabled. What to change • In each agent's AgentSession, set RoomInputOptions / roomInputOptions to include ParticipantKind.AGENT in allowed kinds. This lets an agent detect/transcribe speech coming from another agent. Code examples • Python

from livekit import agents, rtc

session = agents.voice.AgentSession(
  # other settings...
  room_input_options=agents.RoomInputOptions(
    allowed_kinds=[rtc.ParticipantKind.PARTICIPANT_KIND_AGENT]
  )
)

• Node.js

import { voice } from '@livekit/agents';
import { rtc } from 'livekit-client';

const session = new voice.AgentSession({
  // other settings...
  roomInputOptions: {
    allowedKinds: [rtc.ParticipantKind.AGENT]
  }
});

Must-haves for agent-to-agent voice • Both agents connected to the same LiveKit room. • Both agents include

ParticipantKind.AGENT

in their allowed kinds. • TTS configured on agents so synthesized audio is produced and published. • Agents publish their audio tracks to the room (so the other agent can hear them). Useful scenarios • Automated testing (one agent simulates a user, the other responds). • Multi-agent pipelines where one agent hands off to another. Do you want a full example for Node or Python that wires STT → LLM → TTS between two local agents, or do you need help enabling TTS/audio publishing in your current agent? Which runtime are you using? Sources: How to Enable Agent-to-Agent Speech Detection in LiveKit | How to test your agent using another agent | Agents and handoffs | LiveKit docs | Agents Framework | livekit/livekit_composite | DeepWiki | Turns overview | LiveKit docs