Agent wellbeing system — quiet hours, morning/evening nudges, weekly check-ins. Advisory boundaries for AI agents and their humans.
Skill Claude CodeCodex
Detect silent agent failures and memory degradation. Use when asked for an agent health check, when agent behavior seems off or output quality drops, after long unattended runs, or as part of a scheduled maintenance pass.
Skill Claude CodeCodex
Gate outbound notifications through the human's quiet hours. Use before sending any message, notification, nudge, or alert to the human, and when deciding whether something is urgent enough to interrupt them outside working hours.