A Run is the full execution record of one workflow invocation. Every function call, retry, checkpoint, tool invocation, and LLM request is captured in order — with inputs, outputs, timing, and state transitions — automatically, with no instrumentation to add.

Runs are the ground truth for debugging. When something goes wrong in production, you open the Run, scrub through the execution, and see exactly what the system saw. No log correlation, no sampling, no gaps.

You’ll find Runs under Observe → Runs in Studio.

Runs exist as a side effect of how AGNT5 works. The same journal that lets workflows resume from checkpoints is the journal that makes Runs inspectable — there’s no second system to keep in sync.

A deeper guide is in progress. For the execution model behind Runs, see Workflows and Functions.