KubeBuddy Radar MCP Server

🚧 Work in Progress - This MCP server is currently in active development. Features and tools are being tested and refined.

Model Context Protocol server for KubeBuddy Radar - Query 286+ CNCF projects, release notes, security tracking, and your personal dashboard with AI agents like Claude Desktop, Cursor, or Windsurf.

What is this?

This TypeScript MCP server wraps the KubeBuddy Radar API so AI agents can:

• Search and discover CNCF projects
• Get latest release information with AI-parsed notes
• Track security vulnerabilities and CVEs
• Filter by category, CNCF status, and security releases
• Access your personal subscriptions and dashboard

Public tools are routed through Cloudflare cache for speed, while private tools use WordPress Application Passwords for authenticated access.

Installation

Prerequisites

• Node.js 18+ and npm
• A KubeBuddy Radar account (free tier available at radar.kubebuddy.io)

⚠️ Windows Users: Direct installation from GitHub (npm install -g github:KubeDeckio/kubebuddy-radar-mcp) may fail with EPERM errors or missing TypeScript compiler. Use the local build method below.

Option 1: Install from Local Build (Recommended)

git clone https://github.com/KubeDeckio/kubebuddy-radar-mcp.git
cd kubebuddy-radar-mcp
npm install
npm run build
npm pack
npm install -g kubebuddy-radar-mcp-0.0.1.tgz

Option 2: Install from npm (when published)

npm install -g @kubedeck/kubebuddy-mcp-server

Configuration

Claude Desktop

Edit your Claude Desktop config file:

• Windows: %APPDATA%\Claude\claude_desktop_config.json
• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "kubebuddy-radar": {
      "command": "node",
      "args": [
        "C:\\Users\\YOUR_USERNAME\\AppData\\Roaming\\npm\\node_modules\\@kubedeck\\kubebuddy-mcp-server\\dist\\index.js"
      ]
    }
  }
}

Note: Replace YOUR_USERNAME with your Windows username. On macOS/Linux, the path will typically be /usr/local/lib/node_modules/@kubedeck/kubebuddy-mcp-server/dist/index.js.

Restart Claude Desktop, and you're ready to ask questions like:

• "What's the latest Kubernetes version?"
• "Show me recent security releases"
• "What are the most popular CNCF projects?"

If you prefer not to install globally, use npx to launch the MCP server on demand:

{
  "mcpServers": {
    "kubebuddy-radar": {
      "command": "npx",
      "args": [
        "-y",
        "@kubedeck/kubebuddy-mcp-server"
      ]
    }
  }
}

npx -y will download and run the scoped package automatically whenever Claude starts.

Other AI Assistants

MCP support is expanding. Check the official MCP documentation for integration with:

• Cursor IDE
• Windsurf
• VS Code with GitHub Copilot (experimental)

Available Tools

Public Tools (No Authentication)

Private Tools (Authentication Required)

Authentication

Private endpoints require your WordPress username and an Application Password:

1. Go to Account → Developer API access
2. Create an Application Password
3. Use format: username:app_password when prompted by AI agent

Example Queries

Once configured, ask your AI assistant:

Security & Vulnerabilities:

• "What are the latest security releases?"
• "Show me critical CVEs from this week"
• "Are there any security issues in the latest Kubernetes?"

Version & Release Info:

• "What's the latest ArgoCD version?"
• "Show me recent CI/CD tool releases"
• "What breaking changes are in the new Prometheus release?"

Discovery:

• "What are the CNCF graduated projects?"
• "Show me popular service mesh tools"
• "Find projects in the Networking category"

Features

• 🔍 Smart Search - Server-side search with exact match prioritization
• 🏷️ Category Filtering - Filter by CI/CD, Networking, Storage, Security, Observability, etc.
• 🛡️ Security Tracking - CVE lists, severity levels, and security-only filtering
• 🎓 CNCF Maturity Filtering - Filter by graduated/incubating/sandbox status
• 🤖 AI-Parsed Release Notes - Every release includes AI-generated summaries
• ⚡ Cloudflare Cached - Public endpoints are cached for instant responses
• 🔐 Secure Authentication - WordPress Application Passwords for private data

Documentation

• Full Documentation
• API Reference
• Testing Guide
• UX Improvements

Development

# Install dependencies
npm install

# Build TypeScript
npm run build

# Run in development mode
npm run dev

# Test with MCP Inspector
npx @modelcontextprotocol/inspector node dist/index.js

Releases

A GitHub Action named kubebuddy-radar-mcp/.github/workflows/mcp-release.yml now builds, packs, and publishes the MCP server whenever someone publishes a release or triggers the workflow manually.

1. The workflow runs npm ci, npm run build, npm pack, and uploads the generated kubebuddy-radar-mcp-*.tgz artifact.
2. It publishes to https://registry.npmjs.org under @kubedeck/kubebuddy-radar-mcp using the NPM_TOKEN secret, so make sure the repository secret is set before releasing.
3. Trigger it manually with the Run workflow button on GitHub or by creating/publishing a release; the tarball will also be available as an artifact for download.

Update the version field in package.json and anywhere the tarball name is referenced before releasing so npm pack produces the correct filename.

Troubleshooting

Installation fails with EPERM errors? This is a Windows-specific issue. Use the local build method: clone the repo, run npm install && npm run build && npm pack, then npm install -g kubebuddy-radar-mcp-0.0.1.tgz.

Server not responding? Verify the path in your config matches where npm installed the package (typically %APPDATA%\Roaming\npm\node_modules\@kubedeck\kubebuddy-mcp-server\dist\index.js on Windows). Restart your AI assistant after configuration changes.

Schema validation errors? Make sure you're on the latest version. Run npm run build and restart your AI assistant.

Wrong project IDs? Search results now explicitly show IDs in format "ProjectName (ID: 651)".

Need help? Check the testing guide or open an issue on GitHub.

License

MIT

Links

• KubeBuddy Radar
• MCP Documentation
• Report Issues