How-toIntermediateadmin, member
Monitoring Agent Health
Monitor agent availability, response latencies, success rates, and token efficiency from unified health dashboards.
4 min read7,150 viewsUpdated 2026-09-11
Operational health dashboard
Navigate to Agent Detail → Health in the web dashboard or open the agent card in the mobile app to inspect real-time performance indicators:
- Uptime Percentage: Continuous availability score over the rolling 30-day window (target: > 99.8%)
- Average Turn Latency: End-to-end processing time per user interaction or skill tool call (target: < 2.5s)
- Task Success Rate: Percentage of completed workflows without human escalation or unhandled failures
- Token Efficiency: Average token burn per completed objective, helping identify prompt bloat or overly verbose tool loops
Overall Health: 99.9% | Avg Latency: 2.1s | Success Rate: 98.6%
The automated weekly executive digest highlights agents experiencing unusual latency spikes or rapid token consumption shifts.
Configuring proactive alert rules
Set proactive threshold alerts under Settings → Notifications:
| Trigger Condition | Recommended Threshold | Delivery Channels |
|---|---|---|
| High Error Velocity | > 3 errors within 15 minutes | Email, Slack, Mobile Push |
| Token Pool Exhaustion | 80% and 95% pool consumption | Email, Webhook |
| Unresolved Human Approval | Gate pending > 30 minutes | WhatsApp, Slack, SMS |
| Skill Authorization Expiry | 3 days before token expires | Email, Dashboard Banner |
Notifications can be routed to team inboxes, enterprise Slack channels, or mobile push notifications.
monitoringalertsdashboardhealth
Was this helpful?
More questions? Contact support