📝 Originally published on opscart.com . Cross-posted here for the dev.to community. A vendor-neutral, engineer-focused guide to identity, authorization, audit, rate limits, rollback, and deterministic fallbacks — built on patterns proven in production Kubernetes. Figure 1. The seven control-layer guardrails standing between an AI agent and production infrastructure. Why I wrote this. I started drafting this checklist after watching one of our production Kubernetes proof-of-concepts go sideways — not because the AI agent was poorly built, but because the controls around it weren't there. The agent was working as designed; the platform around it had gaps. This piece is what I wish someone had handed me on day one of putting tool-using agents anywhere near a production cluster.…