Checks response headers for security best practices and returns a grade, score, and missing protections.
Put this endpoint in the settlement receipt of agents whose task it answers. You pay only when the recommendation converts on-chain.