Skip to content

Inspect runs

Open Runs in a project to inspect that project’s agent and mechanical workflow activity. Open it in a workspace to see activity across projects.

Use Runtime, Status, Started and Blueprint ID to narrow the list. Open the run ID on the row you want to inspect.

Run detail shows status, start and end times, durations, the pinned definition and version, and a timeline. Waiting isn’t a failed state. It means the run is still active and waiting for its next event.

Copy the full Run ID and Trace ID from Correlation identifiers when reporting a problem to support. Include the time and the visible failure.

For step-by-step workflow diagnosis, follow Test and fix a workflow. Inspect the recorded run’s version before changing the current draft. Editing the draft doesn’t change the configuration a historical run used.

A failed run needs a cause, not another blind attempt. Read its failure and step details, correct the cause, then test the changed workflow.

The run inspector doesn’t yet provide the complete input, output and error payload for every failed or retried attempt. Don’t infer those values from a current draft or a later successful attempt.

Conversation-hosted mechanical workflow history uses the message-retention policy below. This is separate from the high-level Runs catalog.

Data Platform default Free plan Support can change it
Conversation-hosted mechanical workflow history, before pruning 365 days 30 days Yes

Organization overrides can change these values. Don’t treat a run list’s date filter as a retention policy.