Forem: kkb

Introducing Kubernetes MCP: Safe, Read-Only Kubernetes Operations with AI

kkb — Tue, 24 Jun 2025 12:43:16 +0000

Managing Kubernetes clusters is complex. Engineers spend countless hours running kubectl commands, analyzing logs, and debugging issues. What if an AI assistant could help you inspect your cluster, analyze problems, and provide insights—without any risk of accidental modifications?

The kubernetes-mcp project solves this exact challenge.

kubernetes-mcp is an open-source project that bridges Kubernetes clusters with AI assistants through the Model Context Protocol (MCP), ensuring 100% read-only safety.

Get started with kubernetes-mcp →

What is MCP?

Model Context Protocol (MCP) is an open protocol that enables AI assistants to safely interact with external systems through well-defined tools. Think of it as an API specifically designed for AI-human-system collaboration.

What is kubernetes-mcp?

kubernetes-mcp is a specialized MCP server built from the ground up for safe, read-only Kubernetes operations. Using Go and official Kubernetes client libraries, it gives AI assistants comprehensive visibility into your cluster while completely preventing any modifications.

Why Read-Only Matters

Traditional Kubernetes tools come with inherent risks when integrated with AI. A misinterpreted command could delete critical resources or modify production configurations. Kubernetes MCP eliminates these risks entirely:

🔒 Pure read-only architecture: No write operations exist in the codebase
🛡️ Production-safe by design: Impossible to accidentally modify resources
✅ Zero modification risk: Perfect for AI-assisted operations
🧠 AI-native integration: Purpose-built for intelligent assistants

Quick Start

Getting started with kubernetes-mcp takes just two steps:

# Install kubernetes-mcp
go install github.com/kkb0318/kubernetes-mcp@latest

Then configure your MCP-compatible AI assistant:

{
  "mcpServers": {
    "kubernetes": {
      "command": "/path/to/kubernetes-mcp"
    }
  }
}

Four Powerful Tools for AI-Driven Insights

kubernetes-mcp provides four specialized tools that enable AI assistants to intelligently explore and analyze your cluster:

1. `list_resources` - Smart Resource Discovery

Discover resources across your entire cluster with intelligent filtering.

2. `describe_resource` - Deep Resource Analysis

Get detailed information about any resource, including configurations, status, and relationships. AI can analyze complex setups and identify misconfigurations.

3. `get_pod_logs` - Intelligent Log Analysis

Retrieve logs with time-based filtering, allowing AI to correlate events, identify patterns, and diagnose issues across multiple pods.

4. `list_events` - Event-Driven Debugging

Access cluster events to understand what happened and when. AI can reconstruct incident timelines and identify root causes.

Real-World Impact

The beauty of AI-powered exploration is its flexibility. While every AI model behaves differently, here's what we've observed in practice:

Example: GitOps Health Check

You ask: "What GitOps resources are failing in my cluster?"

In our experience, the AI typically:

Discovers all FluxCD/ArgoCD resources across namespaces
Analyzes their current health status
Identifies specific failures and their root causes
Provides actionable remediation steps

What would normally require multiple kubectl commands and deep expertise happens instantly through natural language.

Example: Debugging a Failed Service

You ask: "Check the payment service status"

The AI might investigate by:

Finding pods with "payment" in their names
Checking their current state
Scanning logs for ERROR or FATAL messages
Looking for related warning events

Based on the available information, the AI could identify issues like OOMKilled events, connection timeouts, or configuration errors - helping you quickly focus on the actual problems.

Who Benefits?

DevOps Engineers: Streamline debugging and reduce mean time to resolution
- "What pods are consuming the most memory?"
- "Show me all failing deployments in the last hour"
Developers: Understand deployments and troubleshoot issues without Kubernetes expertise
- "Why is my app not starting?"
- "What environment variables are set for my service?"
Platform Teams: Build AI-assisted operations tools with confidence
- "Generate a report of all deprecated API usage"
- "List all resources without resource limits"
On-Call Engineers: Get instant insights during incidents
- "What changed in the cluster in the last 30 minutes?"
- "Show me error patterns across all services"

Start Today

kubernetes-mcp is open source, production-ready, and designed for seamless integration with any MCP-compatible AI assistant. Experience the future of safe, intelligent Kubernetes operations.

Get started with kubernetes-mcp →

What's Next?

kubernetes-mcp is under active development with new read-only tools being added regularly. We're building a comprehensive suite of AI-powered Kubernetes inspection capabilities, driven by real-world needs and community feedback.

Join the Journey

🌟 Star the project to stay updated
💡 Open issues to suggest features
🤝 Contribute to the development
💬 Share your use cases

Introducing a Custom Operator for Unified Management of Kubernetes Tools

kkb — Sun, 05 May 2024 08:15:00 +0000

Summary

I developed a new custom operator, the Kubernetes Operator Manager (KOM), to simplify the management of Helm packages and other resources in Kubernetes environments and to enable declarative management.
This operator centralizes the deployment and management of various tools, enhancing the efficiency of version control and state monitoring.

https://github.com/kkb0318/kom

Introduction

The Kubernetes ecosystem includes a wide range of tools to meet various workloads and requirements.
These tools are commonly deployed using commands like helm install or kubectl apply.
For example, when using multiple tools such as prometheus and cert-manager in a Kubernetes cluster, you need to execute specific commands for each tool, as follows:

Installation example for cert-manager:

helm repo add jetstack https://charts.jetstack.io --force-update
helm repo update
helm install cert-manager jetstack/cert-manager \
--namespace cert-manager \
--create-namespace \
--version v1.14.5 \
--set installCRDs=true

Installation example for prometheus:

helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
helm repo update
helm install prometheus prometheus-community/prometheus

However, this method has several issues:

It increases operational load as you need to execute commands each time a new tool is introduced.
The installation process is not declarative, making version control and state management challenging.

These issues can lead to confusion, especially when managing multiple Helm packages.
To handle these problems, we developed the Kubernetes Operator Manager (KOM). With KOM, you can manage tool versions declaratively.
This makes it easy to keep track of current versions and upgrade as necessary, leading to more efficient and effective management of Kubernetes clusters.

How KOM Works

KOM operates within kubernetes environments where Flux or ArgoCD is installed.

Managing packages with Flux or ArgoCD requires creating various Custom Resources (CRs).

Flux:

Flux uses CRDs such as HelmRelease and HelmRepository to manage Helm packages. For deploying resources from a Git repository, GitRepository and Kustomization CRDs are used.

ArgoCD
In the case of ArgoCD, resources primarily used include Application CRDs and Secret.

KOM centrally manages these resources, enabling more straightforward and efficient operations. The following diagram outlines KOM's behavior, showing how it manages Flux or ArgoCD's CRs required to deploy Helm and Git resources.

By leveraging the functionalities provided by Flux and ArgoCD, KOM functions as a highly lightweight operator.

How to Use KOM

Installation procedures and basic usage instructions for KOM are detailed in the README, but here we present a more specific usage example.

If you want to install two Helm packages, cert-manager and prometheus, you can configure KOM as follows. This configuration allows you to specify argo or flux depending on the environment.

apiVersion: kom.kkb0318.github.io/v1alpha1
kind: OperatorManager
metadata:
  name: kom
  namespace: kom-system
spec:
  tool: argo # Set to "flux" if using Flux
  cleanup: true
  resource:
    helm:
      - name: jetstack
        url: https://charts.jetstack.io
        charts:
          - name: cert-manager
            version: v1.14.4
            values:
              installCRDs: true
      - name: prometheus-community
        url: https://prometheus-community.github.io/helm-charts
        charts:
          - name: prometheus
            version: 25.20.1

This configuration file uses the specified Helm charts to deploy the necessary packages to your Kubernetes cluster.
By changing the tool setting, you can choose to use either Flux or ArgoCD.
Additionally, setting the cleanup flag automates resource deletion, simplifying environment maintenance.

Conclusion

I developed KOM to manage packages in Kubernetes environments simply and declaratively.

I hope this article has helped you understand KOM's concept, structure, and usage. KOM is still under development, and its API may change, so please refer to the Git repository for the latest information.