Checks whether an AI has permission to access or act on something.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.