Kubernetes AI Agent Directives

Kubernetes is where most infrastructure risk actually lives, because the cluster is the layer that runs everything and yet audits itself least. These Kubernetes directives give AI coding agents a precise, read-only job on a live cluster: detect drift between Git and live state, audit RBAC and network policies, tune autoscaling, diagnose crash loops, and lint manifests before they ship. Each cursor rules kubernetes directive is written to be pasted straight into your rules or instructions file and executed with the real kubectl tools an agent already has. The value is consistency: a platform engineer can hand the same kubernetes audit to an agent in Cursor, Windsurf, or Claude Code and get the same structured report every time. The directives favor narrow, safe operations, so they are suitable for teams that want automation without giving an agent destructive power over a cluster. For anyone searching for cursor rules kubernetes, windsurf rules kubernetes, or ai agent devops directives, this category is the practical starting point: concrete prompts, real commands, and outputs you can review before they influence anything. These are the prompts to keep in one place when the ask is a defensible, repeatable cluster audit. Run them on a cadence and a cluster becomes something an agent can audit on demand, with output a reviewer can trust and a history it can point back to whenever the drift reappears.