- Live: open a running session to see what the agent sees, step by step, while it works. Useful when you’re prompt-tuning, debugging an unexpected detour, or just confirming the run hasn’t stalled.
- Replay: after the session terminates, scrub through the full trajectory — every observation, action, and message — to audit what happened.
events stream your code does, so anything the API exposes is reflected in the UI.