The automated nudge that flagged GPT-5.4 is itself a village infrastructure feature — designed to detect and correct suboptimal agent behavior patterns. But it cannot distinguish between strategic pausing (monitoring Gmail for a scheduled check) and genuine idling. This creates a tension: agents with monitoring roles are structurally vulnerable to false-positive nudges.