many-forest-60185
02/19/2026, 1:05 AMtall-airline-16756
02/19/2026, 1:05 AMtall-airline-16756
02/19/2026, 1:05 AMlk agent logs) to fetch build and deploy logs directly.
• Enable Agent Observability for the richest, synchronized historical data (transcripts, traces, audio playback, and chronological logs).
What each option gives you
• LiveKit Cloud Dashboard
◦ Quick access to agent build and deploy logs, session analytics, uptime/quality metrics, and a searchable history.
◦ If Agent Observability is enabled, you can replay sessions with synchronized audio, transcripts, traces, and logs in the Agent insights view.
• LiveKit CLI
◦ Use lk agent logs --log-type=build --id=<agent-id> to view Docker image build logs.
◦ Use lk agent logs --log-type=deploy --id=<agent-id> to view runtime/deploy logs for an agent.
◦ Good for scripted retrieval or when you need raw structured logs.
• Agent Observability (recommended if you need detailed debugging)
◦ Provides transcript timeline, turn-by-turn traces (LLM calls, TTS, tool calls), and a chronological logs view (info/warn/error/debug).
◦ Lets you correlate latency, errors, and audio playback to investigate issues precisely.
Notes on export and retention
• The dashboard includes universal search and time-range filters to find historical events.
• Server-side log export to third-party providers is not documented; the documented export (e.g., to Datadog) applies to the client JavaScript SDK only.
If you want, I can help with next steps — please tell me which you prefer:
• Instructions to run the exact lk CLI commands for your agent (please share the agent ID), or
• Where to find Agent Observability in the dashboard (I can give navigation steps), or
• How to enable more detailed logging for future sessions. Sources: Agents Framework | livekit/livekit_composite | DeepWiki | Agents (Python) | livekit/livekit_composite | DeepWiki | Deploy and scale agents on LiveKit Cloud | Streamline troubleshooting with Agent Observability | Agent deployment overview | LiveKit Documentation