The failures that matter first are an action identified in a failed run’s history and a connection displaying a warning or error. The failing action requires its root cause to be fixed, while the connection may need to be fixed or reauthenticated before the flow is used again.
How to check the failure point
| Observed signal | What to check | Corrective step |
|---|---|---|
| A flow has failed | Open its run history and identify the failing action. | Fix the root cause before turning the flow back on. |
| A connection shows a warning or error icon | Select the affected connection to determine whether its connection needs attention. | Use the available Fix connection or Re-authenticate option. |
These checks distinguish a specific failed action from a connection problem. A workflow should not be considered ready merely because one symptom disappears; the underlying interruption still needs to be addressed.
What the tester must confirm separately
The cited guidance does not establish whether generated output meets the required quality, whether data and access controls are appropriate, or whether monitoring, cost controls, and ongoing failure handling are adequate. Those points require separate confirmation against the workflow’s intended use.
In particular, clearing a connection warning or correcting a failing action does not by itself prove that the connected AI workflow produces an acceptable result.