You cannot improve what you cannot see
An agent runs unattended by design. The dashboard is how you know what it did: which agents are online, what they are doing right now, and what it cost.
Live status
Each agent shows online while its heartbeat is fresh and offline when it stops. A device that drops off is the first thing to check when a job does not run.
Streaming logs and traces
- The live log streams every event as it happens.
- The deep-dive trace shows the full plan and tool calls.
- The spend tile shows cost per agent and per day.
The audit trail is the backstop
For anything you were not watching live, the hash-chained audit trail is the record. Every action, decision, and reason is there, provable.
A quick daily ritual
- Check the spend tile.
- Scan the log for blocked actions.
- Open any failed run’s trace.
- Confirm every device is online.