can we set maximum number of participants two, one client and once agent, when we creating the room??

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

so this

max_participants

is agent number or both agent and client number??

Good question —

max_participants

counts the total number of participants in the room (agent + client), not per role. Setting

max_participants: 2

limits the room to two people in total. If you need one reserved slot for an agent and one for a client (e.g., prevent two clients from joining), tell me which SDK you’re using and I can suggest a simple server-side check or role-based join flow.

i am using javascript sdk for creatign rooma nd joing the client

Thanks — I’ll show a concise JavaScript example for creating a room with

max_participants: 2

and enforce one agent + one client when joining. 1) Create the room server-side (RoomServiceClient example)

import { RoomServiceClient } from 'livekit-server-sdk'; // adjust import for your package

const svc = new RoomServiceClient(apiKey, apiSecret, wsUrl);

await svc.createRoom({
  name: 'my-room',
  max_participants: 2, // total participants (agent + client)
});

2) Issue join tokens with roles (server-side). Give the agent a token with

metadata

or a

name

that identifies the role, and the client a different token.

import { AccessToken, VideoGrant } from 'livekit-server-sdk';

function createToken(identity: string, room: string, isAgent: boolean) {
  const at = new AccessToken(apiKey, apiSecret, { identity });
  at.addGrant(new VideoGrant({ room }));
  // mark role in token metadata (string)
  at.metadata = isAgent ? 'role=agent' : 'role=client';
  return at.toJwt();
}

// example:
const agentToken = createToken('agent-123', 'my-room', true);
const clientToken = createToken('client-456', 'my-room', false);

3) Enforce role logic when clients attempt to join (recommended server-side check) • When a user requests a join token or attempts to connect, check current participants in the room (via RoomServiceClient or server-side tracking). • Allow join only if: ◦ If user is agent: there is 0 or 1 participant and no existing agent present. ◦ If user is client: there is 0 or 1 participant and no existing client present. • Reject issuing a token or reject the join if rules would be violated. Simple server-side join check (pseudo-TS):

const room = await svc.getRoom({ room: 'my-room' });
const participants = room.participants || [];

const hasAgent = participants.some(p => p.metadata === 'role=agent' || p.identity.startsWith('agent-'));
const hasClient = participants.some(p => p.metadata === 'role=client' || p.identity.startsWith('client-'));

if (isAgent && hasAgent) throw new Error('Agent already connected');
if (!isAgent && hasClient) throw new Error('Client already connected');
if (participants.length >= 2) throw new Error('Room is full');

Notes: •

max_participants: 2

prevents more than two total participants but does not enforce role distribution by itself. • Storing role in token

metadata

or in your own server-side state lets you reliably enforce one-agent/one-client rules. • I can provide a complete runnable example for issuing tokens and checking room participants in Node.js if you want. Which part would you like next?