Kubesel (kubeconfig selector) is your modern approach to working with kubectl configuration in a multi-cluster, multi-namespace environment. Quickly and easily change your active kubectl context, namespace, and cluster through a single program.

Designed from the ground up using the KUBECONFIG environment variable, kubesel creates and manages a unique kubeconfig file for each shell session. If you change your cluster in one pane, it won't affect the other pane where you were planning on running kubectl delete.

Table of Contents:

• Installation
  • Setup
• Features
• Usage
• Tips
  • List Output Formats
  • Adding Kubeconfig Files From a Directory
• Alternatives

go install github.com/eth-p/kubesel

nix profile install github:eth-p/kubesel

For kubesel to set contexts/namespaces per shell, it needs to be able to generate and manage a kubeconfig file associated with specific instances of the shell. To do this, kubesel init creates a shell script that will generate the file and update the KUBECONFIG environment variable.

While you can run the command manually, it is highly recommended to set up your .profile/.rc files to automatically run it whenever a new interactive shell is created. Doing it this way eliminates the risk of multiple shells accidentally sharing the same kubesel-managed kubeconfig file.

if [[ $- = *i* ]]; then
    source <(kubesel init bash)
fi

source <(kubesel init zsh)

if status is-interactive
    kubesel init fish | source
end

• Per-shell context (cluster, user, and namespace).
• Supports bash, zsh, and fish.
• Shell completions.
• Fuzzy matching contexts/clusters/users/namespaces.
• A fzf interface for picking contexts/clusters/users/namespaces.
• Preserves OIDC authentication refresh tokens.
• Shell-scripting friendly list subcommand.
• Manual pages.
• Fancy ANSI colors! (optional)

Change Cluster, User, or Namespace:

kubesel cluster my-cluster      # use this cluster
kubesel user my-user            # use this user
kubesel namespace my-namespace  # use this namespace

Change Cluster, User, and Namespace:

# Use the cluster, user, and namespace from this context.
kubesel context my-context
kubesel context my-context -n # keep the current namespace

View Contexts, Clusters, Users, or Namespaces:

kubesel list clusters
kubesel list contexts
kubesel list users
kubesel list namespaces

The kubesel list command supports changing its output format with --output.
Supported formats are:

• list for just the names in an unsorted list
• table for a table
• col for columns
• col=COL1,COL2 for specific columns
• col=* for all columns

If you keep each cluster in a different kubeconfig file, you can have kubesel automatically include them during initialization:

ls ~/.kube/configs
# prod.yaml
# staging.yaml
# dev.yaml
# kind-myapp.yaml

# Use the `--add-kubeconfigs` flag to include them with a glob pattern.
kubesel init --add-kubeconfigs='~/.kube/configs/*.yaml'

https://github.com/ahmetb/kubectx

• ✅ Shell completions.
• ✅ Fuzzy-finding.
• ⚠️ Changes affect all shells.

https://github.com/Ramilito/kubesess

• ✅ Per-shell cluster/namespace/context.
• ✅ Fuzzy-finding.
• ⚠️ Does not support OIDC refresh tokens.

https://github.com/eth-p/fish-kubeswitch

• ✅ Per-shell cluster/namespace/context.
• ✅ Shell completions.
• ⚠️ Only supports fish shell.
• ⚠️ Wraps kubectl as a shell function.

• no_init_completions
  Removes shell completion loading from the kubesel init scripts.