Super
Equips AI agents with guarded system-administration capabilities using approvals, threat inspection, and append-only audit logging.
Super is an enterprise-ready superuser agent protocol that enables AI models to perform real system administration tasks safely. Operating through a unified choke point, it wraps machine-level execution and GitHub operations with strict safety controls, including dangerous command denylists, human-in-the-loop approval thresholds, and append-only audit logging. Super natively integrates with Cisco AI Defense for runtime threat and data leakage inspection, Cisco Duo Agentic Identity for policy authorization, and Stripe for usage-based metering, transforming autonomous root access into a secure, verifiable, and shippable workflow.