Introduction

Completely uninstalling Minikube on Windows usually means more than removing one Chocolatey package. Cluster profiles, kubeconfig entries, cached files, and driver-specific artifacts can survive the uninstall and interfere with the next local Kubernetes setup.

Delete the Cluster First

Before removing the package, cleanly remove any Minikube cluster state.

minikube status
minikube stop
minikube delete --all --purge

The --purge flag matters because it removes profile data that would otherwise stay under the user profile. If you skip this step and uninstall only the executable, the machine can look clean while still carrying Minikube-specific contexts and cached state.

Uninstall the Chocolatey Package

Once the cluster has been deleted, remove the package itself.

choco list --local-only | findstr /I minikube
choco uninstall minikube -y

Then verify that Chocolatey no longer reports it:

choco list --local-only | findstr /I minikube

That removes the managed package, but not necessarily every related file or configuration entry.

Clean User-Level Directories Carefully

Minikube commonly stores state under the user profile. If the machine is dedicated to local development and no other Kubernetes setup depends on those directories, remove the leftovers.

Remove-Item -Recurse -Force "$env:USERPROFILE\.minikube" -ErrorAction SilentlyContinue
Remove-Item -Recurse -Force "$env:USERPROFILE\.kube" -ErrorAction SilentlyContinue

Be careful with .kube. If you use other clusters on the same machine, do not delete the whole directory. Instead, remove only Minikube-specific entries.

1kubectl config get-contexts
2kubectl config delete-context minikube
3kubectl config delete-cluster minikube
4kubectl config unset users.minikube

That is the safer option on shared development laptops.

Remove Driver-Specific Artifacts

How much extra cleanup is needed depends on the driver that Minikube used.

If the Docker driver was involved, look for leftover containers and networks:

docker ps -a --format "{{.Names}}" | findstr /I minikube
docker network ls --format "{{.Name}}" | findstr /I minikube

If any Minikube-specific artifacts remain, remove only those:

docker rm -f minikube 2>$null
docker network rm minikube 2>$null

If Hyper-V was the driver, check for a leftover VM:

Get-VM | Where-Object { $_.Name -like "*minikube*" }

If one remains and you are sure it belongs to Minikube:

Stop-VM -Name minikube -Force
Remove-VM -Name minikube -Force

Run those commands carefully so you do not delete an unrelated local VM.

Check PATH and Residual Binaries

After the package is gone, confirm there is no stale minikube.exe still on the PATH.

where minikube
where kubectl

If where minikube still prints a path, remove that leftover binary and restart the shell. Old terminals can keep resolving a stale path until they are reopened.

This is a common reason an uninstall appears incomplete even though the package manager step succeeded.

Verify That the Machine Is Clean

A final check should confirm both command resolution and Kubernetes configuration state.

minikube status
kubectl config get-contexts

A clean result generally means one of these outcomes:

• 'minikube is no longer found'
• Minikube reports no profile
• there is no minikube context in kubeconfig

That is the condition you want before reinstalling Minikube or switching to another local Kubernetes option.

Common Pitfalls

The most common mistake is uninstalling the Chocolatey package and assuming everything else vanished with it. Package removal is only one piece of the cleanup.

Another common issue is deleting the entire .kube directory on a machine that also talks to remote clusters. Developers also often forget to clean Docker, Hyper-V, or other driver leftovers and then misdiagnose later Kubernetes problems as fresh installation bugs.

Summary

• Delete Minikube profiles with minikube delete --all --purge before removing the package.
• Uninstall the Chocolatey package separately from state cleanup.
• Remove .minikube and only remove .kube fully if no other clusters depend on it.
• Clean up Docker or Hyper-V leftovers when the old driver created them.
• Verify both PATH resolution and kubeconfig state before calling the uninstall complete.