Skip to main content
Radar runs as a single binary. Point it at a kubeconfig and it serves a web UI on localhost:9280. Use it as a kubectl plugin, a standalone CLI, or a desktop app.

​
CLI

Install script:
Homebrew:
Krew (kubectl plugin):
After install: kubectl radar (as a plugin) or kubectl-radar (standalone). The install script and Scoop lay down the binary as kubectl-radar; the bare radar command works only if you’ve added a symlink yourself or installed via Homebrew.

​
Desktop app

Same UI, native app for macOS / Linux / Windows.

​
Run it

Radar reads your default kubeconfig, opens http://localhost:9280 in your browser, and starts watching the cluster. Common variants:
See CLI flags for the full list.

​
Run it in a cluster

The same binary runs in-cluster as a long-running pod. This is how you give a team shared access to one cluster, and it’s the same pod Radar Cloud installs when you connect a cluster. See In-cluster deployment for the Helm chart.

​
Verify the install

Radar checks for updates when you open the UI (at most once an hour): an anonymous request to releases.skyhook.io/radar/latest, falling back to api.github.com, that prints a notice if a newer version is available. It carries only the version and OS / arch. Anonymous usage stats are sent only if you opt in. Block egress and Radar runs fully offline.

​
Uninstall

The CLI keeps a small config under ~/.radar/. Delete that directory to wipe local preferences and the timeline database.

​
Next

Configuration files

Where Radar stores config, kubeconfig precedence, multi-context.

In-cluster deployment

Run Radar as a shared pod with auth and ingress.

Connect to Cloud

One URL for all your clusters, with team sign-in, roles and audit. Radar still runs in each cluster.