I have one kind of agent working, and would like to have another, created using a different agent_name. Should that go in a separate Python file? Does the filename matter?

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

If I have two agents, in separate Python files, how should I structure the folders where I keep them? I would like to be able to create/dispatch/test each type of agent separately.

Thanks — I’m responding to your question about folder structure for two agents in separate Python files and how to create/dispatch/test each independently. Recommended layout (modular and testable)

project/
├── agents/
│   ├── __init__.py
│   ├── agent_server.py          # shared AgentServer / server instance
│   ├── intake/
│   │   ├── __init__.py
│   │   ├── agent.py             # @server.rtc_session(agent_name="intake-agent")
│   │   └── test.py              # local/console tests for intake agent
│   └── billing/
│       ├── __init__.py
│       ├── agent.py             # @server.rtc_session(agent_name="billing-agent")
│       └── test.py              # local/console tests for billing agent
├── dispatch_rules.yaml         # optional: mapping agent_name -> rules
└── pyproject.toml / requirements.txt

Key points and practices • Use one shared

AgentServer

(or server instance) defined in

agent_server.py

and import it into each

agent.py

. Agents are identified by their

agent_name

in

@server.rtc_session(...)

, not by filename. • Keep each agent’s logic and tests inside its own folder to allow independent development and clear responsibilities. • For local testing, run the per-agent test script or use console mode to exercise the agent without LiveKit dispatch. • For dispatching, specify the exact

agent_name

when creating tokens or routing so jobs go to the correct agent handler. • Use env vars or CLI flags in

test.py

to control which agent runs in dev, and deploy multiple workers with the same

agent_name

for production scale. If you want, share your current repo layout or one agent’s

agent.py

and I’ll suggest a concrete refactor and a sample

test.py

you can run locally. Sources: Build Your First AI Voice Agent in Python: Complete Tutorial | LiveKit | Agents and handoffs | LiveKit Documentation | Avoiding crosstalk when multiple developers share a LiveKit project | LiveK... | Avoiding crosstalk when multiple developers share a LiveKit project | LiveK... | Agents Framework | livekit/livekit_composite | DeepWiki, with interactive elements