Skip to main content

🤖 AI-Native Orchestration

LaraKube CLI is the first Kubernetes orchestrator designed for the age of AI. We provide a Dual-MCP architecture that gives AI agents both local execution power and global fleet visibility.

🔌 Model Context Protocol (MCP)​

LaraKube CLI implements two distinct MCP servers to provide AI agents (like Gemini CLI, Claude Desktop, or Cursor) with the perfect balance of context.

🛠 1. The Local Mechanic (larakube-cli)​

The CLI MCP handles Local Execution. It has direct access to your source code, .env files, and orchestration verbs.

  • Status: 🟢 Ready (6 Expert Tools)
  • Best For: Scaffolding new projects, adding services, and patching local configuration.
  • Tools:
    • inspect-local-code: Deep-scan your project for architectural DNA.
    • local-health-check: Verify Docker, K3d, and host networking status.
    • orchestrate-verb: Run core verbs like up, down, heal, and add.
    • patch-blueprint: Surgically modify .larakube.json.
    • patch-local-env: Update your local .env configuration.
    • run-proxy: Execute artisan, composer, or npm inside the cluster.

One-Click Registration​

You can automatically register the Local Mechanic with your favorite AI tools:

# Register with Gemini CLI
larakube mcp:register --gemini

# Register with Claude Desktop
larakube mcp:register --claude

# Register with all supported tools
larakube mcp:register --all

🧠 2. The Master Architect (larakube-console)​

The Console MCP handles Global Observability. It lives inside your cluster and has access to real-time logs, events, and project history.

  • Status: 🟢 Ready (9 Fleet Tools)
  • Best For: Debugging pod failures, analyzing fleet health, and historical audit trails.
  • Transport: SSE (Server-Sent Events) via your console URL.
  • Tools:
    • get-fleet-status: Bird's-eye view of all registered projects.
    • list-pods: Real-time health check of all project containers.
    • get-project-logs: Fetch the latest logs for any project pod.
    • diagnose-pod: Combined log and status analysis for failing services.
    • get-cluster-events: Check for Kubernetes warnings and failures.
    • explain-architecture: AI analysis of a project's infrastructure.
    • get-project-config: Read the architectural DNA from the cluster.
    • fetch-audit-trail: Historical activity logs for projects or the entire fleet.
    • search-documentation: Real-time RAG search of the LaraKube CLI docs.

Setup Guide​

To add the Master Architect to your AI tool, use the SSE Transport pointing to your Console URL:

{
  "mcpServers": {
    "larakube-console": {
      "url": "https://console.kube/mcp"
    }
  }
}

💡 Pro Tip: Which one should I use?​

  • Use the Local Mechanic (CLI) when you need the AI to do things (build, up, add, change code).
  • Use the Master Architect (Console) when you need the AI to understand things (why is it failing? what happened yesterday?).