#72821
5
58.80
A primitive that lets humans define bounded authority for AI agents, agents execute within it, and anyone verify compliance.