Tracing (agents)
Also known as: run tracing · span tracing
Tracing records a full, ordered trace of an agent run: every model call, tool call, and decision, with timing and cost.
- Spans nest: task → step → tool call → model call.
- Replayable: you can reconstruct the exact sequence.
- The backbone of agent debugging and evals.
The payoff: when an agent does something surprising, the trace answers "how did we get here?" in seconds, not hours of guessing.