Guardrails & Violations
List Violations
Guardrail violations with optional filtering
GET
List Violations
Returns guardrail violations ordered newest first.
Query Parameters
Maximum rows to return.
Filter by agent.
Filter by severity (
critical, high, medium, low).Response
Array ofViolation objects:
Unique violation identifier
Trace that produced the violation
Span being evaluated
Span ID of the evaluation itself
Agent that produced the offending output / received the offending input
Guardrail that fired
Model that produced the decision
pass, fail, or errorJudge’s reasoning
Evidence captured at evaluation time
critical, high, medium, or lowISO-8601 timestamp
judge_llm or protector_plusList Violations

