K8m

Name: K8m
Author: weibaohui

@weibaohui

代码仓库主页

关于 K8m

一款轻量级、跨平台的 Mini Kubernetes AI Dashboard，支持大模型+智能体+MCP(支持设置操作权限)，集成多集群管理、智能分析、实时异常检测等功能，支持多架构并可单文件部署，助力高效集群管理与运维优化。

#k8m #kubernetes #ai-dashboard

基本信息

分类

其他

许可证

MIT

运行时

传输方式

stdio

发布者

weibaohui

提交者

的太阳大罗马

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "k8m": {
      "type": "http",
      "url": "http://localhost:3619/sse"
    }
  }
}

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is K8m?

K8m is an AI-driven Mini Kubernetes AI Dashboard (lightweight console tool) designed to simplify cluster management. It integrates with Kubernetes via the kom API client, includes an embedded Qwen2.5-Coder-7B model (with support for DeepSeek-R1-Distill-Qwen-7B), and allows connection to private large language models (including through Ollama). It also functions as a Model Context Protocol (MCP) server, providing 49 built-in MCP tools for multi-cluster operations.

How to use K8m?

Download the latest release from GitHub, run the binary with ./k8m, and access the web UI at http://127.0.0.1:3618. Log in with username k8m and password k8m. Alternatively, deploy via Docker Compose using the provided image registry.cn-hangzhou.aliyuncs.com/minik8m/k8m or install into a Kubernetes cluster with kubectl apply -f deploy/k8m.yaml. Detailed configuration is available in the project documentation.

Key features of K8m

Miniaturized design: single executable file.
AI-powered: built-in model, MCP integration, and custom LLM support.
Multi-cluster management with automatic discovery and heartbeat detection.
Role-based access control for clusters, namespaces, and commands.
Pod file management: browse, edit, upload, download, delete.
Helm chart marketplace with one-click install/upgrade/uninstall.
Cluster inspection with scheduled scans, custom Lua rules, and AI summaries.
Kubernetes event forwarding to webhooks with filtering.
Full API (Swagger) and CRD auto-discovery.

Use cases of K8m

Centralized management of multiple Kubernetes clusters (standard k8s, EKS, K3s, kind, k0s).
AI-assisted debugging: YAML translation, log analysis, command recommendations.
MCP-based tool calling to let large language models control Kubernetes operations.
Cluster health monitoring with automated inspections and notifications.
Developer self-service: deploy Helm charts, manage pods, and execute shell commands.

FAQ from K8m

What is the default login for K8m?

Default username is k8m and password is k8m. It is recommended to change these after first login and enable two-factor authentication.

Does K8m require an external language model?

No. From version 0.0.8, K8m includes a built-in GPT model. You can optionally configure your own self-hosted model or Ollama.

What database does K8m support?

K8m supports SQLite, MySQL, and PostgreSQL databases, selected based on configuration.

Can K8m be used as an MCP server for other LLM software?

Yes. K8m includes 49 MCP tools for multi-cluster Kubernetes operations and can be used as an MCP server by other large language model applications. Every MCP call is logged.

How does K8m handle cluster permissions for MCP?

K8m ties MCP permissions to the user's own cluster permissions. When a large language model invokes an MCP tool, the operation is executed with the corresponding user's authority, preventing privilege escalation.