K8m

K8M is an AI-driven lightweight Kubernetes console tool that integrates large model capabilities, supports multi-cluster management, and provides MCP services.

Developer tools Virtualization #K8S management #AI-driven #Multi-cluster #MCP service .Go

rating : 3.5 points

downloads : 10.0K

update time : 2025-04-28

Open Site

What is K8M?

K8M is an intelligent management tool designed specifically for Kubernetes. With built-in AI capabilities and a visual interface, it enables non-professional users to easily manage container clusters. It integrates core functions such as resource monitoring, fault diagnosis, and command execution.

How to use K8M?

Simply download and run the single-file program, and access the console through a browser. It supports automatic discovery of multiple clusters via kubeconfig and provides user permission management and operation auditing functions.

Applicable scenarios

Scenarios such as development and testing environment management, teaching demonstrations, operation and maintenance of small and medium-sized production clusters, and AI-assisted fault troubleshooting. It is particularly suitable for teams that need to lower the threshold of using K8s.

Main features

MCP tool integration

It has 49 built-in multi-cluster management tools, supports combining into hundreds of operation instructions, and can be used as an MCP Server for other AI systems to call.

AI operation and maintenance assistant

It integrates large models such as Qwen/DeepSeek and provides intelligent functions such as log analysis, command recommendation, and configuration interpretation.

Multi-cluster management

It automatically scans the kubeconfig directory and centrally manages resources of multiple K8s clusters.

Permission integration

MCP operations automatically inherit users' K8s permissions to ensure security and compliance.

Advantages

Ready to use: Deployed with a single binary file without complex dependencies

Intelligent operation and maintenance: AI automatically diagnoses common problems, reducing the learning cost

Permission security: MCP operations strictly follow the RBAC permission system

Lightweight and efficient: Written in Golang with extremely low resource consumption

Limitations

Limited performance in large-scale clusters (it is recommended to manage less than 50 nodes)

Some advanced functions require configuring the large model API

Support for the ARM architecture is still being improved

How to use

Download and install

Download the executable file for your system from the GitHub Releases page

Start the service

Start the service via the command line, which listens on port 3618 by default

Access the console

Access http://localhost:3618 in your browser and log in with the default account

Usage examples

Intelligent log analysis

When a Pod enters the CrashLoopBackOff state, AI automatically analyzes the logs and provides possible causes

Batch operations on multiple clusters

Use the MCP tool to scale up the nginx Deployment in all clusters simultaneously

Frequently Asked Questions

How to reset the administrator password?

Which large models are supported?

Related resources

Online demonstration

Experience account: demo/demo

Development documentation

Details of architecture design and technical implementation

Docker image

Official container image

🚀 k8m

k8m is a tool that can be run in a containerized or Kubernetes cluster environment, with configurable parameters and environment variables.

🚀 Quick Start

This section provides detailed installation and configuration guides for running the k8m tool, including containerized and Kubernetes cluster deployment methods, as well as environment variable configuration.

📦 Installation

Containerized Deployment

Running with Docker

docker run -it --rm weibaohui/k8m:latest

Parameter Explanation

--help: Display help information.
@<arg>: Specify the path to the configuration file.
-alsologflag: Output to both the log file and the console.
-logtostderr: Output only to standard error.
-v <value>: Set the log level. Valid values are 0 (no logs), 1 (errors), 2 (warnings), 3 (information), 4 (debugging).
--skip_headers: Skip returning header information.
--skip_log_headers: Skip log header information.
-config string: Specify the path to the configuration file.

Running in a Kubernetes Cluster

Using KinD

Installation Steps

brew install kind
kind create cluster --name k8sgpt-demo

Deploying k8m to the Cluster

####### Installation Script

kubectl apply -f https://raw.githubusercontent.com/weibaohui/k8m/refs/heads/main/deploy/k8m.yaml

####### Access Method

By default, NodePort is used for exposure. Please access port 31999. Alternatively, you can configure Ingress yourself.

Configuration Modification Suggestions

It is recommended to modify the configuration by setting environment variables. For example, add env parameters in the deploy.yaml file.

💻 Usage Examples

Environment Variable Explanation

The following environment variables can be set when running the application, for example:

export PORT=8080
export OPENAI_API_KEY="your-api-key"
export GIN_MODE="release"
./k8m

⚠️ Important Note

Environment variables will be overridden by startup parameters.

📚 Documentation

Contact Information

If you have any further questions or need additional assistance, please feel free to contact me!

WeChat (The Sun of the Great Rome): Search for ID daluomadetaiyang and note k8m.

WeChat Group

📄 License

Acknowledgments

zhaomingcheng01: Provided many high - quality suggestions, making outstanding contributions to the usability of k8m.
La0jin: Provided online resources and maintenance, greatly enhancing the display effect of k8m.
eryajf: Provided very useful GitHub Actions, adding automated release, build, and publishing functions to k8m.