Agent Monitors

Agent monitors observe AI agents — their spans, traces, and conversations. All four name an agent source the same way (see each page); they differ in what they alert on.

YAML key	Name	What it monitors
`agent_metric`	Agent Metric Monitor	Statistical metrics over agent spans (null rate, latency, …), ML or fixed thresholds.
`agent_evaluation`	Agent Evaluation Monitor	LLM-as-judge evaluation scores, with eval transforms and sampling.
`agent_trajectory`	Agent Trajectory Monitor	Span occurrence and relationship rules over an agent's trajectory.
`agent_validation`	Agent Validation Monitor	Row-level validation rules over agent spans.