The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard Kubernetes management tools (kubectl and helm) to perform its tasks. These commands are necessary for the skill's primary function of uninstalling a Kubernetes operator and are invoked following a defined safety protocol.
[SAFE]: The skill implements a robust safety contract that mandates a read-only inventory of the cluster state before any destructive actions are proposed. It requires the agent to present exact commands to the user and obtain explicit confirmation before proceeding with deletions, mitigating the risk of accidental or unauthorized resource removal.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the Kubernetes API (such as resource names and status). While this represents a potential surface for indirect prompt injection if the cluster contains maliciously named resources, the risk is minimized by the skill's requirement for human-in-the-loop confirmation and its focus on standard system output rather than arbitrary external content.

nim-operator-uninstall