Explains agent decisions with reasoning chains, confidence scores, and alternative paths.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.