Analyzes input to identify and classify the underlying problem type, returning structured diagnostic labels and confidence scores.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.