When you work with too many kubernetes clusters, you need a way to switch between connecting to each one.
KUBECONFIG is the environment variable that points to your config files,
separated by :,
and each config file can have multiple contexts (cluster, user, namespace combinations),
and one current context.
kubectl has a very basic interface:
kubectl config get-contexts, kubectl config use-context.
It's too much to type,
it's not easy to find a context to connect to,
and it's global to your user.
kubectx
was the first main innovation in this space,
primarily through fuzzy searching with fzf.
But it was still global.
kubeswitch
came next,
allowing isolation between terminal sessions by creating temporary kubeconfig files,
and using shell functions that modify the current environment.
kubie
sounds like it could be better, no need for shell functions.
But it has a critical flaw:
it doesn't properly preserve history.
I have zsh setup to incrementally add and share history between sessions,
but even without that,
kubie would lose history if i had multiple sessions open (the only reason i'd use it).
So, these days, I just use kubeswitch.