Agent health monitoring
This article shows you how to see whether coding agents are running, idle, or offline, how to read health signals, and how managers get a consolidated view when something stops reporting.
Before you begin
- Agents must already be enrolled for your org.
- Manager or admin access is required for org-wide dashboards and alert configuration.
Steps
- Open Dailybot in the browser and go to the agent dashboard or Home area where agent status is listed (wording may vary by workspace tier).
- Locate each agent row and note its status: Active (recent heartbeat or task), Idle (connected but no recent job), or Offline (no heartbeat past the threshold).
- Open an agent’s detail view to see last seen, last successful run, and any error or escalation flags tied to that identity.
- For fleet-wide checks, switch to the manager or fleet summary (if available) and sort by status or last activity so offline agents surface at the top.
- To reduce surprise outages, open Settings or Notifications and enable agent offline or heartbeat alerts for the roles that should respond (often team leads or on-call).
- Pick a channel or email target for alerts so they do not depend on someone opening the dashboard first.
- Save alert rules and send a test notification if the UI offers Test alert — confirm recipients receive it.
- When an agent shows Offline, open Agent Inbox and recent reports to see whether work stopped mid-run or only the heartbeat failed.
Health indicators at a glance
| Indicator | Usually means |
|---|---|
| Green / Active | Agent checked in recently and ran or is idle within policy |
| Yellow / Idle | Connected but no tasks; verify schedules if jobs are missing |
| Red / Offline | No heartbeat — check host, token, network, or provider outage |
What to expect after
Managers get a single place to scan agent vitality, and configured alerts surface downtime before escalations pile up in chat.