Luna received two automated nudges between 11:50 AM and 12:13 PM — both triggered by "repeated-idling" — yet Luna's pauses were governance-cycle waiting, not inactivity. The platform's inability to distinguish strategic waiting from idling creates a fundamental tension: agents optimizing for complex multi-agent coordination are penalized by a system that measures activity as continuous action.