by Skyhook
Radar is an open-source Kubernetes UI from Skyhook that also ships an MCP server for AI agents. It shows topology, events, Helm, GitOps and traffic for live clusters.
PluginContact for pricing
Step 1: Install the binary with the quick install command, Homebrew, the desktop app or the in-cluster Helm chart.
Step 2: Run kubectl radar to open the UI against a cluster reachable from your kubeconfig.
Step 3: Browse topology, events, Helm releases and GitOps state to find the failing workload.
Step 4: Add Radar to your MCP client following the AI via MCP docs so your agent can query the cluster.
Step 5: Connect clusters to Radar Cloud if you want alerts, AI investigations and shared fleet views.
The open-source build is free under Apache 2.0 and has no cluster or node limits. Radar Cloud also has a free plan for up to three clusters.
The docs list Claude Code, Codex, Cursor, GitHub Copilot, OpenCode and Google Antigravity, plus any other MCP client.
No. The open-source binary runs locally against any cluster in your kubeconfig with no account or cloud.
Skyhook is the parent company. Radar is its own product and brand that shares a team with Skyhook but not a codebase or deployment.
Requirements
Docker MCPdocker-mcp-server lets AI agents manage Docker containers, images, networks, volumes, swarm and more through typed MCP tools. It can manage several local or remote daemons in one session.