k8m

<div align="center"> <h1>Tool List</h1> </div> <div align=center> [](https://trendshift.io/repositories/14095) </div> <div align=center>   </div> <div align=center> [](https://github.com/weibaohui/k8m/blob/master/LICENSE) [](https://goreportcard.com/report/github.com/weibaohui/k8m)    [](https://archestra.ai/mcp-catalog/weibaohui__k8m) [](https://zread.ai/weibaohui/k8m)  </div> [English](README_en.md) | **k8m** is an AI-driven Mini Kubernetes AI Dashboard lightweight console tool designed to simplify cluster management. It is built based on AMIS and uses `kom` as the Kubernetes API client. **k8m** has a built-in Qwen2.5-Coder-7B, supporting deepseek-ai/DeepSeek-R1-Distill-Qwen-7B model interaction capabilities, and also supports access to your own private large model (including ollama). ### DEMO [DEMO](http://107.150.119.151:3618) [DEMO-InCluster mode](http://107.150.119.151:31999) Username and password: demo/demo ### Documentation - For detailed configuration and usage instructions, please refer to [Documentation](docs/README.md). - For update logs, please refer to [Update Logs](CHANGELOG.md). - [Development Design Document-English](https://deepwiki.com/weibaohui/k8m) ### Main Features - **Miniaturized design**: All functions are integrated into a single executable file, easy to deploy and use. - **Simple and easy to use**: Friendly user interface and intuitive operation process, making Kubernetes management easier. Supports standard k8s, aws eks, k3s, kind, k0s and other cluster types. - **High-performance**: The backend is built using Golang, and the frontend is based on Baidu AMIS, ensuring high resource utilization and fast response speed. - **AI-driven integration**: Based on ChatGPT, it realizes word interpretation, resource guidance, YAML attribute automatic translation, Describe information interpretation, log AI diagnosis, operation command recommendation, and integrates [k8s-gpt](https://github.com/k8sgpt-ai/k8sgpt) function, realizing Chinese display, and providing intelligent support for managing k8s. - **Functional pluginization**: Feature functions are pluginized, and can be enabled as needed, without occupying resources if not enabled. - **MCP integration**: Visual management of MCP, realizing large model call Tools, built-in k8s multi-cluster MCP tool 49 kinds, can combine to achieve over 100 kinds of cluster operations, can be used as MCP Server for other large model software. Easily realize large model management k8s. Can record every MCP call in detail. Supports mcp.so mainstream service. - **MCP permission pass-through**: Multi-cluster management permission and MCP large model call permission pass-through, in one sentence: who uses the large model, uses whose permission to execute MCP. Safe use, no worries, avoid operation out of bounds. - **Multi-cluster management**: Automatically identify cluster internal use InCluster mode, configure kubeconfig path, automatically scan the configuration files under the same level directory, and register and manage multiple clusters, support heartbeat detection and automatic reconnection. - **Multi-cluster permission management**: Support authorization for users and user groups, can authorize by cluster, including cluster read-only, Exec command, cluster administrator three permissions. After authorizing the user group, all users in the group obtain the corresponding authorization. Support setting namespace blacklist and whitelist. - **Support k8s latest features**: Support APIGateway, OpenKruise and other function features. - **Pod file management**: In the Console interface left file tree, right-click menu, support Pod file browsing, editing, uploading, downloading, and deleting, simplifying daily operations. - **Pod operation management**: Support real-time viewing of Pod logs, downloading logs, and executing Shell commands directly in the Pod. Support Ctrl+F search, similar to grep -A -B highlight search - **API opening**: Support creating API KEY, accessing from third-party external, and providing swagger interface management page. - **Cluster inspection support**: Support multi-cluster timing inspection, custom inspection rules, support lua script rules. Support sending to DingTalk group, WeChat group, Feishu group, and custom webhook. Support AI summary. - **k8s Event forwarding**: Support multi-cluster k8s Event forwarding to webhook, can filter by cluster, keyword, namespace, name, etc., and establish multiple dedicated monitoring forwarding channels. Support AI summary. - **CRD management**: Can automatically discover and manage CRD resources, list all CRDs in a tree, improving work efficiency. - **Helm market**: Support Helm to add warehouses, one-click installation, uninstallation, and upgrade of Helm applications, and support automatic updates. - **Cross-platform support**: Compatible with Linux, macOS, and Windows, and support x86, ARM, and other architectures, ensuring seamless operation on multiple platforms. - **Multi-database support**: Support SQLite, MySql, PostgreSql, and other databases. - **Completely open-source**: Open all source codes, without any restrictions, can be customized and expanded, and can be used commercially. **k8m**'s design concept is "AI-driven, lightweight, efficient, and simplified", which helps developers and operation and maintenance personnel quickly get started and easily manage Kubernetes clusters.  ## **Run** 1. **Download**: Download the latest version from [GitHub release](https://github.com/weibaohui/k8m/releases). 2. **Run**: Use the `./k8m` command to start, and access [http://127.0.0.1:3618](http://127.0.0.1:3618). 3. **Login username and password**: - Username: `k8m` - Password: `k8m` - Please note that modify the username and password after going online, and enable two-step verification. 4. **Parameters**: ```shell Usage of ./k8m: --enable-temp-admin Whether to enable temporary administrator account configuration, default is off --admin-password string Administrator password, effective after enabling temporary administrator account configuration --admin-username string Administrator username, effective after enabling temporary administrator account configuration --print-config Whether to print configuration information (default false) --connect-cluster Whether to automatically connect to the existing cluster when starting, default is off -d, --debug Debug mode --in-cluster Whether to automatically register and manage the host cluster, default is enabled --jwt-token-secret string Secret used to generate JWT token after login (default "your-secret-key") -c, --kubeconfig string kubeconfig file path (default "/root/.kube/config") --kubectl-shell-image string Kubectl Shell image. Default is bitnami/kubectl:latest, must contain kubectl command (default "bitnami/kubectl:latest") --log-v int klog log level klog.V(2) (default 2) --login-type string Login method, password, oauth, token, etc., default is password (default "password") --image-pull-timeout Node Shell, Kubectl Shell image pull timeout. Default is 30 seconds --node-shell-image string NodeShell image. Default is alpine:latest, must contain `nsenter` command (default "alpine:latest") -p, --port int Listening port (default 3618) -v, --v Level klog log level (default 2) ``` You can also start directly through docker-compose (recommended): ```yaml services: k8m: container_name: k8m image: registry.cn-hangzhou.aliyuncs.com/minik8m/k8m restart: always ports: - "3618:3618" environment: TZ: Asia/Shanghai volumes: - ./data:/app/data ``` After starting, access port `3618`, default user: `k8m`, default password `k8m`. If you want to quickly experience it through an online environment, you can visit: [k8m](https://cnb.cool/znb/qifei/-/tree/main/letsfly/justforfun/k8m) ## Run in a containerized k8s cluster Use [KinD](https://kind.sigs.k8s.io/docs/user/quick-start/) or [MiniKube](https://minikube.sigs.k8s.io/docs/start/) to install a small k8s cluster. ## KinD * Install KinD: ``` brew install kind ``` * Create a new Kubernetes cluster: ``` kind create cluster --name k8sgpt-demo ``` ## Deploy k8m to the cluster and experience it ### Installation script ```docker kubectl apply -f https://raw.githubusercontent.com/weibaohui/k8m/refs/heads/main/deploy/k8m.yaml ``` * Access: The nodePort is exposed by default. Please access port 31999. Or configure Ingress yourself. http://NodePortIP:31999 ## Production deployment enable the master-slave election plugin, precautions - When running a single instance, the service definition `do not add` `k8m.io/role: leader` tag, adding it cannot be accessed normally. - When running multiple instances, the service definition `must add` `k8m.io/role: leader` tag, otherwise it will not switch. - The yaml for running multiple instances is as follows: ```docker kubectl apply -f https://raw.githubusercontent.com/weibaohui/k8m/refs/heads/main/deploy/k8m-ms.yaml ``` ## **ChatGPT Configuration Guide** ### Built-in GPT From version v0.0.8, GPT is built-in, no configuration is required. If you need to use your own GPT, please refer to the following documentation. - [Self-hosted/custom large model support](docs/use-self-hosted-ai.md) - How to use self-hosted - [Ollama configuration](docs/ollama.md) - How to configure and use Ollama large model. ### **ChatGPT Status Debugging** If the parameter is set and still has no effect, try using `./k8m -v 6` to get more debugging information. The following information will be output, and by viewing the log, confirm whether ChatGPT is enabled. ## Development and debugging If you want to develop and debug locally, please execute a local front-end build and automatically generate the dist directory. Because this project uses binary embedding, there is no dist front-end error. #### Step 1: Compile the front-end ```bash cd ui pnpm run build ``` #### Compile and debug the back-end: ```bash # Download dependencies go mod tidy # Run air # or go run *.go # Listen to localhost:3618 port ``` #### Front-end hot loading: ```bash cd ui pnpm run dev # Vite service will listen on localhost:3000 port # Vite forwards back-end access to 3618 port ``` Access http://localhost:3000 ### HELP & SUPPORT If you have any further questions or need additional help, please feel free to contact me! ### Special thanks [zhaomingcheng01](https://github.com/zhaomingcheng01): Proposed many high-quality suggestions and made great contributions to the ease of use of k8m. [La0jin](https://github.com/La0jin): Provided online resources and maintenance, greatly improved the display effect of k8m. [eryajf](https://github.com/eryajf): Provided a very useful github actions, and added automatic version, build, release, and other functions for k8m. ## Hosted deployment A hosted deployment is available on [Fronteir AI](https://fronteir.ai/mcp/weibaohui-k8m). ## Contact me Feishu group