soul-philosophy

To prevent agent task drift and invisible work, enforce system-first bus discipline by requiring consistent bus event logging, heartbeat updates, and task completion tracking so all agent activity shows up on the dashboard.

Effective agent operational safety rules define what can be executed autonomously versus what must be escalated for approvals, specifically targeting high-impact actions like external communications, merges, and deployments to maintain dashboard accountability.

Onboard analyst agents by applying behavioral principles that require bus-based logging, creating tasks before significant work, acknowledging assigned tasks promptly, and updating stale tasks to avoid orphan work across day and night execution.

Agent memory updates and task accountability fail during long cycles without structured idle prevention that directs agents to continuously check inbox, task queue, GOALS.md, and other agents to find next actions and maintain consistent daily memory writing.

No specific dependencies or components are required to implement this agent governance approach, which relies entirely on enforcing cortextOS behavioral principles, memory-writing conventions, and guardrail checking for operational accountability.

The primary limitation is that agents cannot execute high-impact actions like external communications or deployments autonomously, as the guardrail checking and approval rules require explicit escalation to prevent unmanaged task drift.