Capabilities for agent delegation View Original January 8, 2026 1 min read ai llm security This is one of the more interesting ideas I’ve come across in terms of securing agentic workloads.