Skip to main content
The Traces panel shows what happened during agent runs.

​
Capabilities

  • Inspect agent messages and tool calls.
  • Review child runs and run timing.
  • Understand how an agent reached an answer.
  • Debug agent behavior with run-level evidence.

​
Use it when

  • An agent response is unexpected.
  • A tool call failed or returned surprising data.
  • You need evidence for debugging agent behavior.
  • Runs: Find run history before inspecting traces.
  • Chat: Reproduce or continue the conversation.
  • Tools: Review tools called by agents.
  • Logs: Inspect runtime output around the run.