Monitoring guardrails

After you configure and run your agent, you can monitor how guardrails behave during execution. This helps verify that detections trigger as expected and that configured actions (such as Block or Log) are working correctly.

You can review guardrail activity in the following places:

• Debug panel – When running your agent in debug mode, guardrail detections appear inline with execution steps, showing which guardrail was triggered and why.

• Trace view – Displays a detailed record of triggered guardrails, including the detection type, severity, scope (agent, LLM, or tool), and action taken.

• Agent logs – Runtime logs capture all guardrail events for auditing and troubleshooting, especially when using the Log action type.

Regularly reviewing these outputs helps you fine-tune guardrail thresholds, validate behavior, and ensure your agent continues to meet security and compliance requirements.