Skip to content

MCP server for secure Kubernetes Interaction via kubectl commands. Enables AI assistants like GitHub Copilot to safely interact with K8s clusters with robust validation and security.

License

Notifications You must be signed in to change notification settings

Joelayo/kubectl-go-mcp-server

Repository files navigation

kubectl-go-mcp-server

A Model Context Protocol (MCP) server that provides Kubernetes cluster interaction capabilities through kubectl commands. This server enables MCP-compatible clients (like VS Code with Copilot) to execute kubectl commands and retrieve Kubernetes cluster information safely and securely.

Features

  • Kubernetes Integration: Execute kubectl commands through MCP interface
  • Interactive Command Protection: Prevents execution of interactive commands that could hang
  • Resource Modification Detection: Identifies commands that modify cluster resources
  • Robust Security: Multiple validation layers to prevent command injection and unsafe operations
  • Configurable Kubeconfig: Support for custom kubeconfig paths
  • Standard Go Project Layout: Following Go best practices for maintainability
  • Cobra CLI Integration: Professional command-line interface with subcommands

Architecture

kubectl-go-mcp-server acts as a bridge between MCP clients (like VS Code with Copilot) and Kubernetes clusters through kubectl commands:

VS Code/Copilot → MCP Client → kubectl-go-mcp-server → kubectl → Kubernetes Cluster

Key Components

  • MCP Server: Handles JSON-RPC communication and tool registration
  • kubectl Tool: Validates and executes kubectl commands safely
  • Security Layer: Prevents interactive commands and command injection

For detailed architecture information, see docs/architecture.md.

pkg/
├── types/          # 🔧 Core interfaces and data structures
│   ├── Tool        # Interface for all MCP tools
│   ├── Schema      # JSON schema definitions
│   └── ExecResult  # Command execution results
│
├── kubectl/        # 🎯 kubectl-specific implementation
│   ├── KubectlTool # Main tool implementation
│   ├── Validation  # Command safety checks
│   └── Execution   # kubectl command runner
│
internal/
├── mcp/           # 🌐 MCP protocol implementation
│   ├── Server     # MCP server and protocol handling
│   ├── Tools      # Tool registry and management
│   └── Protocol   # JSON-RPC message handling
│
└── config/        # ⚙️ Configuration management
    ├── Config     # Application configuration
    └── Defaults   # Default settings

Extension Points

The architecture is designed for extensibility:

  1. New Tools: Implement the Tool interface to add new capabilities
  2. Custom Validation: Add validation layers for specific use cases
  3. Protocol Extensions: Extend MCP handling for additional features
  4. Output Formatters: Add custom result processing

Performance Considerations

  • Concurrent Safety: All components are designed for concurrent access
  • Resource Management: Proper cleanup and resource disposal
  • Timeout Handling: Configurable timeouts for all operations
  • Memory Efficiency: Streaming and buffered I/O for large outputs

Installation

Prerequisites

  • Go 1.23 or later
  • kubectl installed and configured
  • Access to a Kubernetes cluster

Build from Source

# Clone the repository
git clone https://github.com/Joelayo/kubectl-go-mcp-server.git
cd kubectl-go-mcp-server

# Build the binary
make build

# Or install directly
make install

Download Binary

Download the latest release from the releases page for your platform.

Usage

Standalone

# Run with default kubeconfig
./kubectl-go-mcp-server

# Run with custom kubeconfig
./kubectl-go-mcp-server --kubeconfig /path/to/kubeconfig

# Show version
./kubectl-go-mcp-server version

VS Code Integration

To use with VS Code and Copilot, add this MCP server to your VS Code settings:

{
  "mcp": {
    "servers": {
      "kubectl-go-mcp-server": {
        "type": "stdio",
        "command": "/path/to/kubectl-go-mcp-server",
        "env": {}
      }
    }
  }
}

For platform-specific installation details, see the examples directory.

Available Tools

The MCP server provides the following tool:

kubectl

Execute kubectl commands with comprehensive validation and safety checks.

Parameters:

  • command (required): The complete kubectl command to execute (including 'kubectl' prefix)
  • modifies_resource (optional): Indicates if the command modifies resources ("yes", "no", "unknown")

Example:

{
  "name": "kubectl",
  "arguments": {
    "command": "kubectl get pods -o json",
    "modifies_resource": "no"
  }
}

Safety Features:

  • Interactive Command Detection: Prevents hanging on interactive commands like kubectl exec -it, kubectl edit, kubectl port-forward
  • Resource Modification Tracking: Automatically detects destructive operations
  • Command Validation: Ensures only valid kubectl commands are executed

Security

This server implements multiple security layers including command validation, injection prevention, and interactive command blocking. For detailed security information, see:

Development

See CONTRIBUTING.md for development setup, workflow, and contribution guidelines.

Quick Start for Developers

# Install dependencies and build
make deps && make build

# Run tests
make test

# Format and lint code  
make fmt && make lint

Troubleshooting

Common Issues

  • Server not responding: Verify kubectl is installed and kubeconfig is accessible
  • Interactive command errors: Use non-interactive alternatives (see Security Overview)
  • Permission denied: Check kubectl permissions and cluster connectivity

For detailed debugging information, the server logs all tool calls, validation results, and errors.

Support

License

This project is licensed under the MIT License - see the LICENSE file for details.

About

MCP server for secure Kubernetes Interaction via kubectl commands. Enables AI assistants like GitHub Copilot to safely interact with K8s clusters with robust validation and security.

Topics

Resources

License

Contributing

Security policy

Stars

Watchers

Forks

Packages

No packages published