Desktop · macOS, Windows, Linux

Ask your clusters
in plain language.

KubeSpade embeds the AI CLI you already use — Claude Code, Gemini CLI, Codex, GitHub Copilot, or Cursor Agent — with tools scoped to clusters connected in the app.

Reads are free. Apply and delete wait for Approve in the UI. Shell kubectl has no credentials on purpose.

Works with
Claude Code Gemini CLI Codex CLI GitHub Copilot Cursor Agent

Your CLI. Our cluster tools.

No hosted model from KubeSpade — we launch the agent installed on your machine and hand it MCP tools for the clusters you already connected.

Pick once

First open shows Claude, Gemini, Codex, Copilot, or Cursor if they’re on your PATH. The choice sticks; change it in Settings anytime.

Connected clusters only

The AI sees profiles open in KubeSpade. There is no live kubeconfig in the AI shell — cluster access is MCP-only.

Writes need Approve

List, get, describe, and logs run freely. Apply and delete open an Approve dialog in the app before anything mutates.

Same workbench

AI shares the bottom dock with pod exec and logs — tabs, not a second floating window.

Still on-device

Credentials stay local. The agent talks to your clusters through KubeSpade — not through a KubeSpade cloud backend.

Desktop today

In-app AI ships on macOS, Windows, and Linux. Install your preferred CLI, then open the sparkle in the sidebar.

What the AI can see

Only clusters currently connected in the app, through KubeSpade MCP tools. The AI shell does not get a live kubeconfig or your ~/.kube/config.

  • List / get / describe / logs without prompts
  • Apply / delete only after Approve
  • Privacy Policy

What you install

Bring your own CLI — we detect claude, gemini, codex, copilot, or agent (Cursor) on PATH. Auth stays with that vendor’s login.

Try it on Desktop

Install your preferred AI CLI, open KubeSpade, and hit the AI button in the sidebar.