kubernetes-ai-dashboard README

The project's README from github.com/shiyi25928988/kubernetes-ai-dashboard. As one markdown file: README.md.

Kubernetes AI Dashboard

K8s 集群管理 + AI 智能助手平台

Java Spring Boot Vue Kubernetes License

🇨🇳 简体中文 | 🇺🇸 English

目录

简介

Kubernetes AI Dashboard 是一个面向运维人员的 Kubernetes 管理平台，集成 AI 智能助手与 MCP 工具链。基于 Fabric8 Kubernetes Client 管理集群全量资源，通过 Spring AI 提供工具调用与智能对话，并内置 A2A（Agent-to-Agent）协议框架。

[!NOTE] 本平台面向运维场景，提供 K8s 集群资源的统一管理界面与 AI 辅助运维能力。

功能特性

技术栈

类别 技术 版本
语言 Java 21
框架 Spring Boot 3.5.8
AI 框架 Spring AI 1.1.4
K8s 客户端 Fabric8 Kubernetes Client 6.9.2
ORM MyBatis Plus 3.5.14
认证 Sa-Token 1.44.0
对象存储 MinIO 8.6.0
数据库 MySQL 8.0+
缓存 Redis 7.0+
前端框架 Vue 3.5
前端构建 Vite 8
UI 组件 Element Plus 2.13

项目结构

<!-- Experimental: if rendering fails, preview on GitHub -->
graph TD
    A[kubernetes-ai-dashboard] --> B[dashboard 后端应用]
    A --> C[dashboard-fe 前端]
    A --> D[common 公共模块]
    A --> E[ai-base AI 基础]
    A --> F[oss 对象存储]
    A --> G[user 用户认证]
    A --> H[search 全文检索]
    A --> I[message 站内消息]

kubernetes-ai-dashboard/
├── dashboard/              # 后端主应用（cc.shiyi.dashboard）
├── dashboard-fe/           # 前端应用（Vue 3）
├── common/                 # 公共模块（AOP 注解、工具类）
├── ai-base/                # AI 基础（ChatClient 池、聊天记忆）
├── oss/                    # 对象存储（MinIO）
├── user/                   # 用户认证（Sa-Token）
├── search/                 # 全文检索（Lucene）
├── message/                # 站内消息
└── pom.xml

快速开始

环境要求

启动依赖服务

cd docker
docker-compose up -d mysql redis minio

后端编译运行

mvn clean install -DskipTests
cd dashboard
mvn spring-boot:run

后端启动于 http://localhost:8080

前端开发

cd dashboard-fe
npm install
npm run dev

前端启动于 http://localhost:3000

配置

配置文件 dashboard/src/main/resources/application.yml，所有值支持 ${ENV_VAR:default} 环境变量覆盖。

环境变量 说明 默认值
SPRING_DATASOURCE_URL MySQL JDBC 地址 jdbc:mysql://localhost:3306/mpi_dashboard
SPRING_DATA_REDIS_HOST Redis 地址 localhost
MINIO_ENDPOINT MinIO 地址 http://localhost:9000
SPRING_AI_OPENAI_BASE_URL AI API 地址 https://api.deepseek.com
SPRING_AI_OPENAI_API_KEY AI API Key <YOUR_API_KEY>
KUBERNETES_CLIENT_CONFIG_URL kubeconfig 下载 URL -
KUBERNETES_CLIENT_CONFIG_PATH kubeconfig 本地路径 -
KUBERNETES_CLIENT_TRUST_CERTS 信任 K8s API 证书 true
A2A_API_KEY A2A 协议 API Key <YOUR_API_KEY>

[!IMPORTANT] Kubernetes 客户端必须显式配置 kubeconfig（KUBERNETES_CLIENT_CONFIG_URL 或 KUBERNETES_CLIENT_CONFIG_PATH），否则启动失败。不支持集群内 ServiceAccount 自动检测。

API 参考

内部管理 API（/api/v1/*，需 Sa-Token 登录）

方法 路径 说明
POST /api/v1/ai/chat AI 对话（流式/非流式）
GET/POST/PUT/DELETE /api/v1/k8s/* K8s 资源管理
GET /api/v1/k8s/cluster/stats 集群统计
GET /api/v1/resources/ownership 资源归属查询
POST/GET /api/v1/user/* 用户管理

A2A 协议（/a2a，API Key 认证）

方法 路径 说明
GET /a2a/card 获取 AgentCard
GET /.well-known/agent-card.json 标准发现端点
POST /a2a JSON-RPC 2.0（message/send、tasks/get、tasks/cancel）
MCP 工具列表（7 个）
工具服务 能力
KubernetesMcpService 命名空间/Pod/Deployment 等资源操作、集群统计
SystemInfoMcpService CPU/内存/磁盘/运行时间
CommandLineMcpService 命令执行（危险命令黑名单）
FileMcpService 文件读写（沙箱限制）
SshMcpService 远程 SSH 命令
OssMcpService MinIO 上传/列表
DockerRegistryMcpService Docker 镜像仓库查询

部署

Docker 部署详见 DOCKER_README.md。

# 构建后端镜像
docker build -f dashboard/Dockerfile -t <REGISTRY>/mpi-dashboard-backend:latest .

# 构建前端镜像
docker build -f dashboard-fe/Dockerfile -t <REGISTRY>/mpi-dashboard-frontend:latest ./dashboard-fe

贡献

欢迎提 Issue 与 Pull Request。

许可证

本项目采用内部许可证（Internal License），未公开开源。

kubernetes-ai-dashboard (shiyi25928988/kubernetes-ai-dashboard)

authentication
Searching Database…

299 ·GitHub
Kubernetes-ai-dashboard
What is this?Kubernetes AI Dashboard is a web platform for ops teams that manages the full set of Kubernetes resources (Namespace, Pod, Deployment, Service, Ingress, PVC, ConfigMap, Secret, Event, Job, CRD) and adds an AI assistant with tool-calling, a built-in MCP toolchain, and an Agent-to-Agent (A2A) protocol framework for AI-assisted cluster operations.
Is this a library I import, or a standalone app?It's a standalone application: a Java 21 / Spring Boot 3.5 backend built on the Fabric8 Kubernetes client, paired with a Vue 3 + Vite frontend, not something you embed in another project.
How do I get it running?Start the supporting services (MySQL, Redis, MinIO) with Docker Compose, then build and run the Spring Boot backend with Maven and the Vue frontend with Node.js; Docker deployment details are in DOCKERREADME.md.
What do I need installed before running it?You need JDK 21+, Maven 3.9+, Node.js 18+, Docker & Docker Compose, and access to a Kubernetes v1.24+ cluster.
Any install pitfalls to watch for?The Kubernetes client must be given an explicit kubeconfig via KUBERNETESCLIENTCONFIGURL or KUBERNETESCLIENTCONFIGPATH or the app fails to start — it does not auto-detect an in-cluster ServiceAccount.
What AI features does it include?Spring AI powers a streaming chat assistant with tool-calling and context memory, backed by 7 built-in MCP tool services covering K8s, system info, command execution, files, SSH, object storage and Docker registry queries.
What is the A2A protocol endpoint for?It exposes a JSON-RPC 2.0 Agent-to-Agent endpoint at /a2a with an AgentCard discovery route, letting external agents send messages and manage tasks (message/send, tasks/get, tasks/cancel); the README notes it as a reserved extension point.
How does resource ownership tracking work?Resources created through the dashboard are tagged with the creating user, and users are only allowed to delete resources they themselves created.
How is login secured?User login uses Sa-Token with 32-character random tokens and Redis-backed sessions, separate from the API-key authentication required for the A2A agent protocol endpoints.
What are similar projects?JeecgBoot, evenfire, casdoor, Observal, spring-ai-alibaba, ruoyi-vue-pro, apisix, rankme.fast
Which topics is it in?kubernetes-cluster, element-plus, springboot, object-storage, a2a, cloud-native, vite, mysql, redis, model-context-protocol, authentication, vue
Creator
shiyi25928988
Size
9.1 MB
Stars
299
Updated
CommitAug 12
Age
2mo
Commits · 90 Days