Build kubectl commands with correct bash or PowerShell quoting and every destructive flag explained. Nothing is executed and no cluster is contacted.
kubectl get · changes nothing
Lists resources, or one resource, in a chosen output format.
Lists resources, or one resource, in a chosen output format.
kubectl get podsNothing here is executed. No kubeconfig is read, no cluster is contacted, and the command is text for you to check and run yourself — so it may well name a resource that does not exist. Read the command before running it, particularly anything marked as destroying something: kubectl delete has no confirmation prompt and no undo.
kubectl Command Builder runs entirely on your device — nothing is executed, no kubeconfig is read, and no cluster is contacted.
You’re on the Free plan. Compare plans
The quoting is the part that earns its keep. A JSONPath expression is full of braces and quotes, and bash and PowerShell treat both completely differently — which is why a kubectl command copied from a blog post works for one colleague and fails for another with an error about an unexpected token.
This quotes for the shell you pick: single quotes in bash, with the close-escape-reopen dance when the value itself contains one, and doubled quotes in PowerShell where braces inside a single-quoted string are literal. Pick the wrong shell and the command is still wrong; pick the right one and it is right.
Destructive flags are offered and explained rather than hidden. `--force` with `--grace-period=0` is sometimes the only thing that works, and alongside it is the part nobody mentions: the pod leaves the API immediately without the kubelet confirming the container stopped, so for a StatefulSet pod the volume can still be attached, the replacement cannot start, and two instances can briefly write to the same storage.
Every command is marked by what it does — changes nothing, changes state, or destroys something — and several flags carry a note explaining why they are there. `--ignore-daemonsets` on a drain is effectively required, because DaemonSet pods are not evictable and the drain refuses to start without it.
Nothing is executed. No kubeconfig is read, no cluster is contacted, and the output is text for you to check and run yourself.
Because PowerShell interprets the braces and quotes differently from bash. Double quotes let it try to expand the expression; single quotes are literal, and an embedded single quote has to be doubled. Pick PowerShell here and the quoting comes out right for it.
Sometimes necessary and never free. It removes the pod from the API without waiting for the kubelet to confirm the container stopped, so the process may still be running. For a StatefulSet pod the volume can remain attached, which blocks the replacement and can briefly allow two instances to write to the same storage. It is the right answer for a pod on a node that is genuinely gone, and the wrong one otherwise.
Because the current container has not started yet. `--previous` reads the instance that died, which is where the reason is. Note the kubelet keeps only one previous instance, so a pod that has restarted twice has already discarded the first failure's logs.
No. It reads no kubeconfig, contacts no cluster and executes nothing. It produces a command string for you to read and run yourself — which also means it may name a resource that does not exist.