Skip to Content

Trace

The Trace lens shows how a published, Active workflow executes. It reuses the same canvas as Build, overlaid with live run state.

Prerequisites

  • Workflow is Active (publish from Build first)
  • For Run now, the published graph needs at least one API trigger

If those are missing, empty-states deep-link you back to Build.

Runs

  • Run list — select a run from history (?run=<runId> focuses one)
  • Auto-picks the active or most recent run when none is focused
  • Live detail polls while a run is active (~2.5s)

Controls

ActionWhen
Run nowActive workflow + API trigger; optional trigger picker if multiple
CancelWhile the run is active (“Cancelling…”)
RetryOnly when the run Failed

Canvas overlay

Nodes and edges reflect execution:

  • Status dots and outlines (running, succeeded, failed, waiting, …)
  • Activity line on the active path
  • Auto-focus the first failed node on a failed run

Timeline and step I/O

Run timeline (left): chronological steps with status and duration; selecting a step focuses its node (two-way sync with the canvas).

Step I/O drawer (progressive disclosure):

  1. Summary — status, timing, function, started / finished timestamps
  2. Condition — for filter / branch steps, re-evaluates CEL against the run context and labels the taken branch
  3. Input / Output — pretty JSON with copy
  4. Run context — accumulated trigger / vars / nodes / signal
  5. Raw execution record

Milestones show progress (for example “3 of 7 done”) with a segmented status bar.

Run states

Typical states surfaced in the UI: Queued, Running, Succeeded, Failed, Cancelled, Waiting approval, Waiting timer.

Trace does not auto-publish Build edits. Always Publish from Build before expecting Run now to exercise your latest graph.

Last updated on