Checks and filters content for policy violations and unsafe material.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.