Prerequisites

Before using tabby-k8s-exec, make sure the following are in place.

1. A working kubeconfig file

The plugin reads a kubeconfig file directly — by default ~/.kube/config, though you can point it at any file in the profile settings. It does not create or modify this file; it must already have at least one context pointing at a cluster you can reach.

If you can already run kubectl get pods successfully against your cluster, your kubeconfig is ready to use with this plugin as-is.

2. Credentials for your kubeconfig’s auth method

The plugin resolves credentials using the same logic kubectl itself uses, so whatever your kubeconfig already relies on continues to work unchanged:

• Static token / client certificate — nothing extra is needed.

• ``exec``-based auth plugins (used by managed clusters such as Amazon EKS, GKE, or any setup using kubelogin) — the referenced binary (aws, gke-gcloud-auth-plugin, kubelogin, …) must be installed and available on PATH, and you must have an active, unexpired session with that provider at the moment you connect. For example, an EKS kubeconfig’s exec entry runs aws eks get-token under the hood — if your AWS CLI session has expired, that command fails, and so does the connection. This is inherent to how exec-based auth works and applies equally to plain kubectl; the plugin cannot work around it.

Note

Once a token has been obtained, it is cached in memory for its stated lifetime (an EKS token is normally valid for about 15 minutes), so the underlying aws/gke-gcloud-auth-plugin/etc. command is not re-run on every action — only when no valid cached token exists yet, or the cached one has expired. Every new connection (opening a new tab) starts with its own cache, though, so it can trigger a fresh credential fetch even if another tab or the settings dialog already has one cached.

Once both are in place, continue to Installation.