How Radar by Skyhook.io Is the Next Step for Kubernetes Dashboards

In my previous article, I compared Headlamp and Rancher in the Kubernetes dashboard space and looked at where each one stands today. If you haven’t read it yet, I’d really appreciate it if you gave it a read :)

After publishing it, Nadav from Skyhook reached out and suggested I try their newly open-sourced tool, Radar.

I set it up, used it properly on a local cluster, and I’ found it seriously impressive. This article is a continuation of that comparison, because there is now a new player that deserves attention.

Setup Experience

I spun Radar up on a single-node Kubernetes cluster locally.

The setup was simple. The documentation is clear, the GitHub repo is well structured, and there are multiple ways to run it. You can use the desktop app, deploy it in-cluster with Helm, or run it locally.

The desktop app did not work for me for some reason, so I installed it using Homebrew:

brew install skyhook-io/tap/radar
kubectl radar

It was running in minutes.

What I like here is that Radar runs locally and connects directly to your kubeconfig. Nothing gets installed into your cluster unless you choose the in-cluster Helm option. That makes it easy to test without worrying about cleanup or leftover resources.

Radar is still early-stage, but it does not feel unfinished.

First Impressions

When you first open Radar, you land on a clean home view that shows the essentials. Pods, deployments, nodes, unhealthy workloads, topology, traffic, timeline, MCP integration. Everything important is visible without digging through menus.

It feels organized.

The topology tab is one of the strongest parts of the experience. Instead of scrolling through lists of resources, you can actually see how things are connected in real time.

Switching between resource relationships and traffic views makes debugging more intuitive.

The resources tab covers what you expect from a Kubernetes dashboard. You can scale workloads, view logs, restart pods, exec into containers, check YAML, and review health status. Nothing feels clunky or half-built.

It is straightforward, which is exactly what you want in a dashboard.

The Standout Feature for Me: MCP Integration (BETA)

I have experimented with integrating AI tools with Kubernetes clusters before, and giving an AI raw kubectl access is messy. In Radar, MCP server is enabled by default and provides minified, structured output. It strips unnecessary metadata, summarizes health, deduplicates warning events, filters logs, redacts secrets, and stays read-only. It also respects cluster RBAC.

If you do not want it enabled, you can turn it off:

radar --no-mcp

This makes Radar one of the most AI-ready dashboards right now. With Kubernetes workflows increasingly moving toward AI-assisted debugging, this feels like a forward-looking decision.

Helm and TLS, Practical Additions

Radar includes built-in Helm management.

You can see releases across namespaces, inspect values, compare revisions, and perform rollbacks or upgrades directly from the UI. In real-world environments, Helm is everywhere. Having this built in removes the need to constantly switch back to the terminal.

In Headlamp, you need to install a plugin for Helm support, which adds another step and extra setup.

Another feature I genuinely appreciate is TLS certificate visibility.

Radar parses TLS secrets and surfaces certificate subject, issuer, and expiry dates. Anyone who has dealt with expired certificates in production knows how painful and avoidable that can be. Having expiry information visible in your dashboard is a very practical feature.

The Limitations

Radar currently lacks fine-grained user-based access control when deployed in-cluster as a shared service.

If you expose it for multiple users, they effectively share the same permission scope as the underlying service account. For larger teams, this can be limiting.

It is already an open issue and being discussed, so it is clearly on the roadmap. For local use or smaller teams, it is not a blocker. For bigger environments, it will matter.

Final Thoughts

Radar brings together dashboard functionality, Helm management, TLS inspection, traffic visualization, and AI integration in a way that feels cohesive.

It is still evolving, but it already feels like a solid product built with real-world use in mind.

If you want to try Radar yourself, you can check it out here: Radar