A readable notation for verified AI-agent loops: goals, context, authority, cycles, verifiers, stop rules, and return artifacts.