Detects or scores human-likeness from the /v1/human endpoint.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.