Rule-based monitoring across the agent fleet — 10 rules, 2 acknowledged in the last 24h.
Alerts
6 of 6 widgets shown
8 of 10 enabled
Success rate dropped
success_rate < 95% over 5m
2m ago
last fired
Latency p95 elevated
p95_latency > 5s over 10m
14m ago
last fired
Token spend spike
spend_rate > $200/hr
58m ago
last fired
Tool error rate
tool_errors / runs > 2%
4h ago
last fired
Timeout cluster
timeouts > 10 in 5m
3h ago
last fired
Eval regression
eval_score < baseline − 0.05
5h ago
last fired
Loop detection
same_tool_calls > 8 in run
18h ago
last fired
Budget at 90%
monthly_spend / budget > 0.9
—
never
Cold start spike
p50_cold_start > 3s
3w ago
last fired
Schema drift
tool_schema_hash changed
—
never
6 in 24h · 4 open
on research-agent-v2
Success rate fell to 87.4% over the last 5 minutes.
COMPOSED OF · 2 GALLERY CARDS