Runs
A run is one execution of an Action.
Run views help you understand what happened, including step-by-step results and failures.
Runs are your primary source of truth when validating behavior in real conditions.
For details on configuring fallback behavior when runs fail, see Error Handling.
What You Can Inspect
- run status,
- start and duration,
- credits consumed (see Credits),
- selected action and trigger context,
- per-step output and errors,
- related metadata.
Why Runs Matter
Run data helps you answer practical questions quickly:
- Did the Action Flow do what we expected?
- Where did the first failure occur?
- Was the issue input quality, step logic, or external dependency?
- Is this a one-off run or a repeating pattern?
Typical Statuses and What They Mean
- Completed: Run finished successfully.
- Failed: A step or dependency failed and the run could not complete.
- Running: Run is currently in progress.
Use status together with step details, not in isolation.
Common Troubleshooting Flow
- Filter runs by Action and status.
- Open a failed run.
- Inspect step logs to find the first failing step.
- Update the Action and test again.
If available, compare with the last successful run to identify what changed (input shape, prompt, step config, or tool response).
Operational Habits That Help
- Keep step names precise and consistent.
- Surface key business identifiers via metadata for easy filtering.
- Review failed runs in batches to spot recurring issues.
- Validate fixes by running multiple representative samples.
Operational Tip
Use consistent step names and structured outputs to speed up debugging.