Technical guide to securing multi-agent AI systems with zero trust principles, delegation chain validation, and behavioral ...