ServiceNow agents can observe, decide, and act across workflows, data, identities, and connected systems. That makes agentic security broader than model safety alone. A secure posture must cover the full lifecycle: design, approval, deployment, runtime control, incident response, evidence, and retirement. The goal is simple: every agent must be known, approved, constrained, observable, and stoppable.