Agent Guardrails — Tiered HITL Governance for AI Agent Writesby Ivan FodjoFreeDeterministic tiered governance and human-in-the-loop guardrails for AI agent write-actions.20