Detects phishing and typosquatted domains by comparing them against 40+ major brands.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.