🔊 echo-agent

A simple test agent that echoes back messages for development and debugging.

Current implementation is text-only

As of now echo-agent only echoes text — it responds with You said: {input} (see agents/echo-agent/src/echo_agent/agent.py). The ECHO_MODE variable and the Audio / Both modes described below are not yet implemented; they document the intended design only.

If you are looking for the round-trip audio test on the system status page, that does not use this agent — it uses a pure LiveKit SFU loopback (publish on a publisher token, subscribe on a listener token in the same room) wired up in frontend-ui/src/components/shared/ReadinessCheck/AudioOutputModal.tsx. No echo-agent pod is involved.

Overview

echo-agent is a minimal agent implementation used for:

• Testing LiveKit connectivity
• Verifying audio pipeline works
• Development and debugging
• Integration testing

How It Works

The echo agent:

1. Receives audio from the LiveKit room
2. Transcribes the audio using STT
3. Echoes the transcription back as text
4. (Optionally) Synthesizes the text as audio

Audio In → STT → Echo Text → (Optional TTS) → Audio Out

Use Cases

Testing LiveKit Connection

Verify that your LiveKit server is properly configured:

# Deploy echo agent
curl -X POST http://localhost:3000/sessions/{sessionId}/agents \
  -H "Content-Type: application/json" \
  -d '{"role": "echo-agent"}'

# Speak into microphone
# Agent should repeat what you said

Verifying Audio Pipeline

Check that audio is being captured and transmitted:

1. Deploy echo-agent
2. Speak clearly
3. Verify you hear your words echoed back
4. Check transcription appears in the UI

Debugging Data Channel

Test data channel messaging:

// Send a test message
const data = {
  type: 'user_text',
  data: 'Hello Echo!'
};
room.localParticipant.publishData(
  new TextEncoder().encode(JSON.stringify(data)),
  { reliable: true }
);

// Echo agent will respond with the same message

Configuration

Minimal configuration required:

Variable	Description	Required
LIVEKIT_URL	LiveKit server URL	Yes
LIVEKIT_API_KEY	LiveKit API key	Yes
LIVEKIT_API_SECRET	LiveKit API secret	Yes
STT_PROVIDER	Speech-to-text provider	Optional
ECHO_MODE	text, audio, or both	Planned — not yet implemented

Echo Modes

note

Only text echo is implemented today. The audio/both modes below describe the intended design and are not wired up yet.

Text Mode (current behaviour)

Echoes the received text straight back as the agent's text response (You said: {input}).

Audio Mode (planned)

Would echo audio back without processing:

• Receives audio stream
• Transmits back as-is
• Useful for latency testing

Both Mode (planned)

Would transcribe, echo text, and synthesize an audio response:

• Full pipeline test
• Most comprehensive verification

Resource Requirements

Resource	Request	Limit
CPU	50m	200m
Memory	128Mi	512Mi

Deployment

Via API

curl -X POST http://localhost:3000/sessions/{sessionId}/agents \
  -H "Content-Type: application/json" \
  -d '{
    "role": "echo-agent",
    "config": {
      "echoMode": "both"
    }
  }'

Via Frontend

1. Create or open a session
2. Click "Deploy Agent"
3. Select "echo-agent" from the dropdown
4. Deploy

Troubleshooting

No Echo Response

1. Check LiveKit connection:

   kubectl logs <echo-agent-pod> -n ai-agents

2. Verify room exists:

   curl http://localhost:7880/rooms

3. Check audio track is published:

   • Open browser DevTools
   • Check Network tab for WebRTC connections

Delayed Echo

1. Check STT service latency
2. Verify network connectivity
3. Check pod resource usage:

   kubectl top pod <echo-agent-pod> -n ai-agents

Garbled Audio

1. Check sample rate settings
2. Verify audio codec compatibility
3. Test with text-only mode first

Development Use

Echo agent is useful during development for:

# Quick connection test
async def test_connection():
    agent = EchoAgent()
    await agent.connect(room_name="test-room")
    # If you hear echo, connection works
    await agent.disconnect()

See Also

• Agents Overview - All agent types
• LiveKit Integration - LiveKit setup
• First Agent - Deployment guide