Every AI agent active in a selfdriven engagement holds a delegated KERI credential — a cryptographic proof of what it is authorised to do, issued by the accountable human conductor. No agent operates outside its bounded scope.
In a selfdriven engagement, no AI agent operates with open-ended access. Every agent in the roster has a defined domain, a scoped authority credential issued via KERI delegation, and a time-limited mandate tied to the specific engagement.
The Agent Roster is the live registry of agents active in your engagement — what each agent is authorised to do, who delegated that authority, and how to verify it independently.
Traditional AI systems rely on configuration files and API keys to define what an agent can access. If those credentials are compromised or misconfigured, the scope of the breach is opaque. selfdriven agents receive authority through KERI delegated inception events — cryptographic, auditable, and revocable.
Every agent in the roster has a verifiable chain of authority from the issuing conductor to the specific task scope, anchored in the Key Event Log.
selfdriven maintains a standing roster of specialist AI agents, each aligned to one or more Areas of Focus. For any engagement, a subset of these agents is delegated and activated by the assigned human conductor.
Every agent in an engagement receives authority through a formal cryptographic delegation — not a configuration file, not an API key, but a verifiable KERI credential chain.
You don't need to trust selfdriven's word that an agent is authorised. The KERI credential chain is independently verifiable by any party with access to the engagement KEL.