Your agent finished successfully — but what if it finished the wrong task?
A user flags a failure mode worse than an outright error: an agent reports success after completing the wrong task, and the success message hides the mistake. The fix discussed is defining what done looks like before the agent starts, in a form checkable afterward — the exact record, file, or confirmation. If the expected outcome cannot be stated in one sentence, the agent is free to complete something else and call it success.