Overview

This page describes things that are common across all skills.

Natural-language arguments

Unless a skill’s page says otherwise, anything you type after the skill name is passed to the agent as a free-form instruction. You don’t need to memorize a grammar, just describe what you want.

> /cluster-status only pods
> /investigate backend server keeps crashing
> /logs errors from the last hour on api pods
> /audit-outdated container images in kube-system

The agent resolves the intent, picks the right resources, and narrows the scope accordingly. Flags (below) still work if you’d rather be explicit — mix and match freely.

K8s Context

By default the skills will use your kubeconfig default context unless you tell them otherwise. You can use natural language to change context and kstack will remember your choice for the rest of your session:

> /cluster-status
[ ... ]
> redo with prod
[ ... ]

Global flags

These flags are supported by every skill:

--context <ctx>

Override the current kubeconfig context

--help

Open the skill’s reference documentation page in your browser

More will land here over time. When an individual skill page documents its own flags, assume the global flags are also available on top.