Builds agent simulation scenarios from task risk, edge cases, tool failures, and review criteria.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.