A curated list of tools, papers, frameworks, and best practices for hardening agentic AI systems — covering prompt injection defense, runtime sandboxing, protocol security, red teaming, and governance standards.