Safety & ops
What is Trace?
A trace is the ordered log of steps in an agent run: inputs, tools, observations, outputs.
Traces are the receipt. They should redact secrets but keep causal order.
Why it matters
A trace is the ordered log of every step in an agent run: the input, each tool call with its parameters and result, the model's decisions, and the final output. Traces are the receipt — they prove what the agent did and why.
Traces are essential for debugging, auditing, and evaluation. Without them, agent behavior is a black box.
Key takeaways
- 1Traces are the ordered log of every agent step.
- 2They should redact secrets but preserve causal order.
- 3Traces enable debugging, auditing, and evaluation.
Common mistakes
- ✕Not redacting secrets (API keys, PII) from traces.
- ✕Dropping failed steps from traces — failures are critical debugging information.
Related terms
Concept neighborhood
Terms linked from Trace in the glossary graph.