Analyzes input and returns structured JSON for AI guardrail checks.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.