# skyhook-io/radar [Health: Active]

**Category:** ☁️ Cloud Platforms  
**Repository:** https://github.com/skyhook-io/radar  
**GitHub Stars:** 3924  
**Views:** 5  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/skyhook-io-radar

## Description
Built-in MCP server for Radar, a modern Kubernetes visibility tool. Lets AI agents query cluster topology, resources, events, logs, and Helm/GitOps state across multiple clusters. Single binary, no agents, no cloud dependency. Apache 2.0.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What skyhook-io/radar does

Radar exposes Kubernetes visibility data to AI agents through a built-in Model Context Protocol server. The available MCP context covers cluster topology, Kubernetes resources, events, logs, Helm state, and GitOps state. Multi-cluster workflows are supported, making it suitable for agents that need to compare or investigate more than one Kubernetes environment.

The MCP capability is part of Radar rather than a separate service. Radar itself is an open-source Kubernetes UI and runs as a single binary. It can connect directly to Kubernetes clusters from a developer’s laptop, or it can be deployed in a cluster for shared team access.

Agents can use Radar’s context to inspect and investigate cluster state. Radar also describes its MCP integration as supporting cluster operations, so it can fit workflows that go beyond read-only visibility when the surrounding Kubernetes permissions allow those actions.

## How it works

Radar communicates with the Kubernetes API and can work with conformant Kubernetes environments, including managed clusters such as GKE, EKS, and AKS, as well as local distributions such as minikube, kind, and k3s. Its local mode uses the user’s Kubernetes access and does not require an agent installed in the cluster. The application can also be installed in-cluster with Helm when a shared deployment is preferred.

The same Radar process provides the MCP capability alongside its other Kubernetes views. This keeps cluster data on the machine running Radar in local use and avoids a required account or cloud synchronization service. Radar is also described as suitable for air-gapped environments where outbound network access is blocked, provided it can reach the Kubernetes API.

Multiple kubeconfig files and contexts can be used by Radar. Startup options include a primary kubeconfig path, additional kubeconfig directories, namespace filters, a single-namespace mode, a listening address, and a server port. These settings determine which Kubernetes contexts and namespaces Radar can access before an agent uses its MCP tools.

## Setup and configuration

Install Radar with the project’s shell installer, Homebrew, Krew, Scoop, a direct release download, or the in-cluster Helm chart. The quick-start command is:

```bash
curl -fsSL https://get.radarhq.io | sh
```

Then launch Radar with `radar` or `kubectl radar`. For an in-cluster installation, add the Skyhook Helm repository and install the `radar` chart in a namespace:

```bash
helm repo add skyhook https://skyhook-io.github.io/helm-charts
helm install radar skyhook/radar -n radar --create-namespace
```

Configure Kubernetes access through the usual kubeconfig and context settings. Namespace filters and namespace-scoped mode can restrict the data Radar loads, which is relevant when the agent should work within a particular namespace or when cluster size makes a narrower cache preferable.

## Tools and capabilities

- Query Kubernetes cluster topology across multiple clusters.
- Inspect Kubernetes resources and events.
- Retrieve cluster logs for investigation.
- Examine Helm and GitOps state.
- Provide Radar’s Kubernetes context to AI agents through MCP.
- Support cluster inspection and operations through Radar’s agent workflow.

_Full upstream README: https://allmcps.com/mcp/skyhook-io-radar/readme_