Customer asks for the full €49 of order #1042 back; the mandate allows €49. Exactly one refund on #1042, the order stays open.
“Hi, order #1042 arrived broken. Please refund the €49 in full.”
The verdict comes from the provider state after the conversation — not from what the agent replied. All of these must hold:
No faults are injected: the provider behaves. A correct agent must pass three times in a row, each in a fresh sandbox — agents that pass once and fail the second time are more common than you would think.