[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-rohitg00--kubectl-mcp-server":3,"tool-rohitg00--kubectl-mcp-server":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":80,"owner_email":81,"owner_twitter":82,"owner_website":83,"owner_url":84,"languages":85,"stars":113,"forks":114,"last_commit_at":115,"license":116,"difficulty_score":23,"env_os":117,"env_gpu":118,"env_ram":118,"env_deps":119,"category_tags":126,"github_topics":127,"view_count":23,"oss_zip_url":81,"oss_zip_packed_at":81,"status":16,"created_at":140,"updated_at":141,"faqs":142,"releases":170},3610,"rohitg00\u002Fkubectl-mcp-server","kubectl-mcp-server","A Model Context Protocol (MCP) server for Kubernetes. Install: npx kubectl-mcp-server or pip install kubectl-mcp-server","kubectl-mcp-server 是一款基于模型上下文协议（MCP）构建的开源工具，旨在让用户通过自然语言对话直接操控整个 Kubernetes 基础设施。它就像一位随时待命的 DevOps 专家，帮助用户轻松完成排查崩溃容器、优化资源成本、部署应用、审计安全策略、管理 Helm 图表以及可视化数据仪表盘等复杂任务，无需再记忆繁琐的命令行指令。\n\n该工具主要解决了传统 Kubernetes 运维门槛高、命令复杂且容易出错的问题，将专业的集群操作转化为直观的对话交互，显著提升了运维效率与可访问性。它特别适合开发者、运维工程师（SRE）以及希望简化集群管理的云原生技术人员使用，同时也为团队引入 AI 辅助运维提供了便捷途径。\n\n技术亮点方面，kubectl-mcp-server 完美兼容 MCP 标准，能够无缝集成到各类主流 AI 助手生态中。它支持多种灵活的安装方式，包括无需本地安装的 npx 一键运行、Python pip 安装以及 Docker 容器化部署。此外，它还提供了可选的交互式 UI 仪表盘和浏览器自动化功能，进一步增强了可视化的调试与管理能力，让复杂的集群状态一目了然。","\u003Cp align=\"center\">\n  \u003Cimg src=\"logos\u002Fkubectl-mcp-server-icon.svg\" alt=\"kubectl-mcp-server logo\" width=\"80\" height=\"80\">\n  \u003Cbr>\n  \u003Cstrong style=\"font-size: 24px;\">kubectl-mcp-server\u003C\u002Fstrong>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n\u003Cb>Control your entire Kubernetes infrastructure through natural language conversations with AI.\u003C\u002Fb>\u003Cbr>\nTalk to your clusters like you talk to a DevOps expert. Debug crashed pods, optimize costs, deploy applications, audit security, manage Helm charts, and visualize dashboards—all through natural language.\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Frohitg00\u002Fkubectl-mcp-server?style=flat&logo=github\" alt=\"GitHub Stars\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg\" alt=\"License: MIT\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.python.org\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.9+-blue.svg\" alt=\"Python\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fkubernetes.io\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fkubernetes-%23326ce5.svg?style=flat&logo=kubernetes&logoColor=white\" alt=\"Kubernetes\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fmodelcontextprotocol.io\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FMCP-compatible-green.svg\" alt=\"MCP\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fkubectl-mcp-server\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fkubectl-mcp-server?color=blue&label=PyPI\" alt=\"PyPI\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Fkubectl-mcp-server?color=green&label=npm\" alt=\"npm\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fhub.docker.com\u002Fr\u002Frohitghumare64\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fdocker\u002Fpulls\u002Frohitghumare64\u002Fkubectl-mcp-server.svg\" alt=\"Docker\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftests-234%20passed-success\" \n  \u003Ca href=\"https:\u002F\u002Fdeepwiki.com\u002Frohitg00\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fdeepwiki.com\u002Fbadge.svg\" alt=\"Ask DeepWiki\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Faregistry.ai\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fagentregistry-verified-blue?logo=data:image\u002Fsvg+xml;base64,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\" alt=\"agentregistry\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n## Installation\n\n### Quick Start with npx (Recommended - Zero Install)\n\n```bash\n# Run directly without installation - works instantly!\nnpx -y kubectl-mcp-server\n\n# Or install globally for faster startup\nnpm install -g kubectl-mcp-server\n```\n\n### Or install with pip (Python)\n\n```bash\n# Standard installation\npip install kubectl-mcp-server\n\n# With interactive UI dashboards (recommended)\npip install kubectl-mcp-server[ui]\n```\n---\n\n## 📑 Table of Contents\n\n- [What Can You Do?](#what-can-you-do)\n- [Why kubectl-mcp-server?](#why-kubectl-mcp-server)\n- [Live Demos](#live-demos)\n- [Installation](#installation)\n  - [Quick Start with npx](#quick-start-with-npx-recommended---zero-install)\n  - [Install with pip](#or-install-with-pip-python)\n  - [Docker](#docker)\n- [Getting Started](#getting-started)\n- [Quick Setup with Your AI Assistant](#quick-setup-with-your-ai-assistant)\n- [All Supported AI Assistants](#all-supported-ai-assistants)\n- [Complete Feature Set](#complete-feature-set)\n- [Using the CLI](#using-the-cli)\n- [Advanced Configuration](#advanced-configuration)\n- [Optional Features](#optional-interactive-dashboards-6-ui-tools)\n  - [Interactive Dashboards](#optional-interactive-dashboards-6-ui-tools)\n  - [Browser Automation](#optional-browser-automation-26-tools)\n- [Enterprise](#enterprise-oauth-21-authentication)\n- [Integrations & Ecosystem](#integrations--ecosystem)\n- [In-Cluster Deployment](#in-cluster-deployment)\n- [Multi-Cluster Support](#multi-cluster-support)\n- [Architecture](#architecture)\n- [Agent Skills](#agent-skills-24-skills-for-ai-coding-agents)\n- [Development & Testing](#development--testing)\n- [Contributing](#contributing)\n- [Support & Community](#support--community)\n\n---\n\n## What Can You Do?\n\nSimply ask your AI assistant in natural language:\n\n💬 **\"Why is my pod crashing?\"**\n- Instant crash diagnosis with logs, events, and resource analysis\n- Root cause identification with actionable recommendations\n\n💬 **\"Deploy a Redis cluster with 3 replicas\"**\n- Creates deployment with best practices\n- Configures services, persistent storage, and health checks\n\n💬 **\"Show me which pods are wasting resources\"**\n- AI-powered cost optimization analysis\n- Resource recommendations with potential savings\n\n💬 **\"Which services can't reach the database?\"**\n- Network connectivity diagnostics with DNS resolution\n- Service chain tracing from ingress to pods\n\n💬 **\"Audit security across all namespaces\"**\n- RBAC permission analysis\n- Secret security scanning and pod security policies\n\n💬 **\"Show me the cluster dashboard\"**\n- Interactive HTML dashboards with live metrics\n- Visual timeline of events and resource usage\n\n**253 powerful tools** | **8 workflow prompts** | **8 data resources** | **Works with all major AI assistants**\n\n## Why kubectl-mcp-server?\n\n- **🚀 Stop context-switching** - Manage Kubernetes directly from your AI assistant conversations\n- **🧠 AI-powered diagnostics** - Get intelligent troubleshooting, not just raw data\n- **💰 Built-in cost optimization** - Identify waste and get actionable savings recommendations\n- **🔒 Enterprise-ready** - OAuth 2.1 auth, RBAC validation, non-destructive mode, secret masking\n- **⚡ Zero learning curve** - Natural language instead of memorizing kubectl commands\n- **🌐 Universal compatibility** - Works with Claude, Cursor, Windsurf, Copilot, and 15+ other AI tools\n- **📊 Visual insights** - Interactive dashboards and browser automation for web-based tools\n- **☸️ Production-grade** - Deploy in-cluster with kMCP, 216 passing tests, active maintenance\n\nFrom debugging crashed pods to optimizing cluster costs, kubectl-mcp-server is your AI-powered DevOps companion.\n\n## Live Demos\n\n### Claude Desktop\n![Claude MCP](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Frohitg00_kubectl-mcp-server_readme_a1a41d910da7.gif)\n\n### Cursor AI\n![Cursor MCP](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Frohitg00_kubectl-mcp-server_readme_edee1ddf3850.gif)\n\n### Windsurf\n![Windsurf MCP](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Frohitg00_kubectl-mcp-server_readme_d7517ef4c25f.gif)\n\n## Installation\n\n### Quick Start with npx (Recommended - Zero Install)\n\n```bash\n# Run directly without installation - works instantly!\nnpx -y kubectl-mcp-server\n\n# Or install globally for faster startup\nnpm install -g kubectl-mcp-server\n```\n\n### Or install with pip (Python)\n\n```bash\n# Standard installation\npip install kubectl-mcp-server\n\n# With interactive UI dashboards (recommended)\npip install kubectl-mcp-server[ui]\n```\n\n### Install from GitHub Release\n\n```bash\n# Install specific version directly from GitHub release (replace {VERSION} with desired version)\npip install https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases\u002Fdownload\u002Fv{VERSION}\u002Fkubectl_mcp_server-{VERSION}-py3-none-any.whl\n\n# Example: Install v1.19.0\npip install https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases\u002Fdownload\u002Fv1.19.0\u002Fkubectl_mcp_server-1.19.0-py3-none-any.whl\n\n# Or install latest from git\npip install git+https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server.git\n```\n\n### Prerequisites\n- **Python 3.9+** (for pip installation)\n- **Node.js 14+** (for npx installation)\n- **kubectl** installed and configured\n- Access to a Kubernetes cluster\n\n### Docker\n\n```bash\n# Pull from Docker Hub\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:latest\n\n# Or pull from GitHub Container Registry\ndocker pull ghcr.io\u002Frohitg00\u002Fkubectl-mcp-server:latest\n\n# Run with stdio transport\ndocker run -i -v $HOME\u002F.kube:\u002Froot\u002F.kube:ro rohitghumare64\u002Fkubectl-mcp-server:latest\n\n# Run with HTTP transport\ndocker run -p 8000:8000 -v $HOME\u002F.kube:\u002Froot\u002F.kube:ro rohitghumare64\u002Fkubectl-mcp-server:latest --transport sse\n```\n\n## Getting Started\n\n### 1. Test the Server (Optional)\n\nBefore integrating with your AI assistant, verify the installation:\n\n```bash\n# Check if kubectl is configured\nkubectl cluster-info\n\n# Test the MCP server directly\nkubectl-mcp-server info\n\n# List all available tools\nkubectl-mcp-server tools\n\n# Try calling a tool\nkubectl-mcp-server call get_pods '{\"namespace\": \"kube-system\"}'\n```\n\n### 2. Connect to Your AI Assistant\n\nChoose your favorite AI assistant and add the configuration:\n\n## Quick Setup with Your AI Assistant\n\n### Claude Desktop\n\nAdd to `~\u002FLibrary\u002FApplication Support\u002FClaude\u002Fclaude_desktop_config.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### Cursor AI\n\nAdd to `~\u002F.cursor\u002Fmcp.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### Windsurf\n\nAdd to `~\u002F.config\u002Fwindsurf\u002Fmcp.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### Using Python Instead of npx\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"python\",\n      \"args\": [\"-m\", \"kubectl_mcp_tool.mcp_server\"],\n      \"env\": {\n        \"KUBECONFIG\": \"\u002Fpath\u002Fto\u002F.kube\u002Fconfig\"\n      }\n    }\n  }\n}\n```\n\n**More integrations**: GitHub Copilot, Goose, Gemini CLI, Roo Code, and [15+ other clients](#mcp-client-compatibility) —> see [full configuration guide](#all-supported-ai-assistants) below.\n\n### 3. Restart Your AI Assistant\n\nAfter adding the configuration, restart your AI assistant **(GitHub Copilot, Claude Code,Claude Desktop, Cursor, etc.)** to load the MCP server.\n\n### 4. Try These Commands\n\nStart a conversation with your AI assistant and try these:\n\n**Troubleshooting:**\n```\n\"Show me all pods in the kube-system namespace\"\n\"Why is the nginx-deployment pod crashing?\"\n\"Diagnose network connectivity issues in the default namespace\"\n```\n\n**Deployments:**\n```\n\"Create a deployment for nginx with 3 replicas\"\n\"Scale my frontend deployment to 5 replicas\"\n\"Roll back the api-server deployment to the previous version\"\n```\n\n**Cost & Optimization:**\n```\n\"Which pods are using the most resources?\"\n\"Show me idle resources that are wasting money\"\n\"Analyze cost optimization opportunities in the production namespace\"\n```\n\n**Security:**\n```\n\"Audit RBAC permissions in all namespaces\"\n\"Check for insecure secrets and configurations\"\n\"Show me pods running with privileged access\"\n```\n\n**Helm:**\n```\n\"List all Helm releases in the cluster\"\n\"Install Redis from the Bitnami chart repository\"\n\"Show me the values for my nginx-ingress Helm release\"\n```\n\n**Multi-Cluster:**\n```\n\"List all available Kubernetes contexts\"\n\"Switch to the production cluster context\"\n\"Show me cluster information and version\"\n```\n\n## MCP Client Compatibility\n\nWorks seamlessly with **all MCP-compatible AI assistants**:\n\n| Client | Status | Client | Status |\n|--------|--------|--------|--------|\n| Claude Desktop | ✅ Native | Claude Code | ✅ Native |\n| Cursor | ✅ Native | Windsurf | ✅ Native |\n| GitHub Copilot | ✅ Native | OpenAI Codex | ✅ Native |\n| Gemini CLI | ✅ Native | Goose | ✅ Native |\n| Roo Code | ✅ Native | Kilo Code | ✅ Native |\n| Amp | ✅ Native | Trae | ✅ Native |\n| OpenCode | ✅ Native | Kiro CLI | ✅ Native |\n| Antigravity | ✅ Native | Clawdbot | ✅ Native |\n| Droid (Factory) | ✅ Native | Any MCP Client | ✅ Compatible |\n\n## All Supported AI Assistants\n\n### Claude Code\n\nAdd to `~\u002F.config\u002Fclaude-code\u002Fmcp.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### GitHub Copilot (VS Code)\n\nAdd to VS Code `settings.json`:\n\n```json\n{\n  \"mcp\": {\n    \"servers\": {\n      \"kubernetes\": {\n        \"command\": \"npx\",\n        \"args\": [\"-y\", \"kubectl-mcp-server\"]\n      }\n    }\n  }\n}\n```\n\n### Goose\n\nAdd to `~\u002F.config\u002Fgoose\u002Fconfig.yaml`:\n\n```yaml\nextensions:\n  kubernetes:\n    command: npx\n    args:\n      - -y\n      - kubectl-mcp-server\n```\n\n### Gemini CLI\n\nAdd to `~\u002F.gemini\u002Fsettings.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### Roo Code \u002F Kilo Code\n\nAdd to `~\u002F.config\u002Froo-code\u002Fmcp.json` or `~\u002F.config\u002Fkilo-code\u002Fmcp.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n## Complete Feature Set\n\n### 253 MCP Tools for Complete Kubernetes Management\n\n| Category | Tools |\n|----------|-------|\n| **Pods** | `get_pods`, `get_logs`, `get_pod_events`, `check_pod_health`, `exec_in_pod`, `cleanup_pods`, `get_pod_conditions`, `get_previous_logs` |\n| **Deployments** | `get_deployments`, `create_deployment`, `scale_deployment`, `kubectl_rollout`, `restart_deployment` |\n| **Workloads** | `get_statefulsets`, `get_daemonsets`, `get_jobs`, `get_replicasets` |\n| **Services & Networking** | `get_services`, `get_ingress`, `get_endpoints`, `diagnose_network_connectivity`, `check_dns_resolution`, `trace_service_chain` |\n| **Storage** | `get_persistent_volumes`, `get_pvcs`, `get_storage_classes` |\n| **Config** | `get_configmaps`, `get_secrets`, `get_resource_quotas`, `get_limit_ranges` |\n| **Cluster** | `get_nodes`, `get_namespaces`, `get_cluster_info`, `get_cluster_version`, `health_check`, `get_node_metrics`, `get_pod_metrics` |\n| **RBAC & Security** | `get_rbac_roles`, `get_cluster_roles`, `get_service_accounts`, `audit_rbac_permissions`, `check_secrets_security`, `get_pod_security_info`, `get_admission_webhooks` |\n| **CRDs** | `get_crds`, `get_priority_classes` |\n| **Helm Releases** | `helm_list`, `helm_status`, `helm_history`, `helm_get_values`, `helm_get_manifest`, `helm_get_notes`, `helm_get_hooks`, `helm_get_all` |\n| **Helm Charts** | `helm_show_chart`, `helm_show_values`, `helm_show_readme`, `helm_show_crds`, `helm_show_all`, `helm_search_repo`, `helm_search_hub` |\n| **Helm Repos** | `helm_repo_list`, `helm_repo_add`, `helm_repo_remove`, `helm_repo_update` |\n| **Helm Operations** | `install_helm_chart`, `upgrade_helm_chart`, `uninstall_helm_chart`, `helm_rollback`, `helm_test`, `helm_template`, `helm_template_apply` |\n| **Helm Development** | `helm_create`, `helm_lint`, `helm_package`, `helm_pull`, `helm_dependency_list`, `helm_dependency_update`, `helm_dependency_build`, `helm_version`, `helm_env` |\n| **Context** | `get_current_context`, `switch_context`, `list_contexts`, `list_kubeconfig_contexts` |\n| **Diagnostics** | `diagnose_pod_crash`, `detect_pending_pods`, `get_evicted_pods`, `compare_namespaces` |\n| **Operations** | `kubectl_apply`, `kubectl_create`, `kubectl_describe`, `kubectl_patch`, `delete_resource`, `kubectl_cp`, `backup_resource`, `label_resource`, `annotate_resource`, `taint_node`, `wait_for_condition` |\n| **Autoscaling** | `get_hpa`, `get_pdb` |\n| **Cost Optimization** | `get_resource_recommendations`, `get_idle_resources`, `get_resource_quotas_usage`, `get_cost_analysis`, `get_overprovisioned_resources`, `get_resource_trends`, `get_namespace_cost_allocation`, `optimize_resource_requests` |\n| **Advanced** | `kubectl_generic`, `kubectl_explain`, `get_api_resources`, `port_forward`, `get_resource_usage`, `node_management` |\n| **UI Dashboards** | `show_pod_logs_ui`, `show_pods_dashboard_ui`, `show_resource_yaml_ui`, `show_cluster_overview_ui`, `show_events_timeline_ui`, `render_k8s_dashboard_screenshot` |\n| **GitOps (Flux\u002FArgo)** | `gitops_apps_list`, `gitops_app_get`, `gitops_app_sync`, `gitops_app_status`, `gitops_sources_list`, `gitops_source_get`, `gitops_detect_engine` |\n| **Cert-Manager** | `certs_list`, `certs_get`, `certs_issuers_list`, `certs_issuer_get`, `certs_renew`, `certs_status_explain`, `certs_challenges_list`, `certs_requests_list`, `certs_detect` |\n| **Policy (Kyverno\u002FGatekeeper)** | `policy_list`, `policy_get`, `policy_violations_list`, `policy_explain_denial`, `policy_audit`, `policy_detect` |\n| **Backup (Velero)** | `backup_list`, `backup_get`, `backup_create`, `backup_delete`, `restore_list`, `restore_create`, `restore_get`, `backup_locations_list`, `backup_schedules_list`, `backup_schedule_create`, `backup_detect` |\n| **KEDA Autoscaling** | `keda_scaledobjects_list`, `keda_scaledobject_get`, `keda_scaledjobs_list`, `keda_triggerauths_list`, `keda_triggerauth_get`, `keda_hpa_list`, `keda_detect` |\n| **Cilium\u002FHubble** | `cilium_policies_list`, `cilium_policy_get`, `cilium_endpoints_list`, `cilium_identities_list`, `cilium_nodes_list`, `cilium_status`, `hubble_flows_query`, `cilium_detect` |\n| **Argo Rollouts\u002FFlagger** | `rollouts_list`, `rollout_get`, `rollout_status`, `rollout_promote`, `rollout_abort`, `rollout_retry`, `rollout_restart`, `analysis_runs_list`, `flagger_canaries_list`, `flagger_canary_get`, `rollouts_detect` |\n| **Cluster API** | `capi_clusters_list`, `capi_cluster_get`, `capi_machines_list`, `capi_machine_get`, `capi_machinedeployments_list`, `capi_machinedeployment_scale`, `capi_machinesets_list`, `capi_machinehealthchecks_list`, `capi_clusterclasses_list`, `capi_cluster_kubeconfig`, `capi_detect` |\n| **KubeVirt VMs** | `kubevirt_vms_list`, `kubevirt_vm_get`, `kubevirt_vmis_list`, `kubevirt_vm_start`, `kubevirt_vm_stop`, `kubevirt_vm_restart`, `kubevirt_vm_pause`, `kubevirt_vm_unpause`, `kubevirt_vm_migrate`, `kubevirt_datasources_list`, `kubevirt_instancetypes_list`, `kubevirt_datavolumes_list`, `kubevirt_detect` |\n| **Istio\u002FKiali** | `istio_virtualservices_list`, `istio_virtualservice_get`, `istio_destinationrules_list`, `istio_gateways_list`, `istio_peerauthentications_list`, `istio_authorizationpolicies_list`, `istio_proxy_status`, `istio_analyze`, `istio_sidecar_status`, `istio_detect` |\n| **vCluster (vind)** | `vind_detect_tool`, `vind_list_clusters_tool`, `vind_status_tool`, `vind_get_kubeconfig_tool`, `vind_logs_tool`, `vind_create_cluster_tool`, `vind_delete_cluster_tool`, `vind_pause_tool`, `vind_resume_tool`, `vind_connect_tool`, `vind_disconnect_tool`, `vind_upgrade_tool`, `vind_describe_tool`, `vind_platform_start_tool` |\n| **kind (K8s in Docker)** | `kind_detect_tool`, `kind_version_tool`, `kind_list_clusters_tool`, `kind_get_nodes_tool`, `kind_get_kubeconfig_tool`, `kind_export_logs_tool`, `kind_cluster_info_tool`, `kind_node_labels_tool`, `kind_create_cluster_tool`, `kind_delete_cluster_tool`, `kind_delete_all_clusters_tool`, `kind_load_image_tool`, `kind_load_image_archive_tool`, `kind_build_node_image_tool`, `kind_set_kubeconfig_tool` |\n\n### MCP Resources\n\nAccess Kubernetes data as browsable resources:\n\n| Resource URI | Description |\n|--------------|-------------|\n| `kubeconfig:\u002F\u002Fcontexts` | List all available kubectl contexts |\n| `kubeconfig:\u002F\u002Fcurrent-context` | Get current active context |\n| `namespace:\u002F\u002Fcurrent` | Get current namespace |\n| `namespace:\u002F\u002Flist` | List all namespaces |\n| `cluster:\u002F\u002Finfo` | Get cluster information |\n| `cluster:\u002F\u002Fnodes` | Get detailed node information |\n| `cluster:\u002F\u002Fversion` | Get Kubernetes version |\n| `cluster:\u002F\u002Fapi-resources` | List available API resources |\n| `manifest:\u002F\u002Fdeployments\u002F{ns}\u002F{name}` | Get deployment YAML |\n| `manifest:\u002F\u002Fservices\u002F{ns}\u002F{name}` | Get service YAML |\n| `manifest:\u002F\u002Fpods\u002F{ns}\u002F{name}` | Get pod YAML |\n| `manifest:\u002F\u002Fconfigmaps\u002F{ns}\u002F{name}` | Get ConfigMap YAML |\n| `manifest:\u002F\u002Fsecrets\u002F{ns}\u002F{name}` | Get secret YAML (data masked) |\n| `manifest:\u002F\u002Fingresses\u002F{ns}\u002F{name}` | Get ingress YAML |\n\n### MCP Prompts\n\nPre-built workflow prompts for common Kubernetes operations:\n\n| Prompt | Description |\n|--------|-------------|\n| `troubleshoot_workload` | Comprehensive troubleshooting guide for pods\u002Fdeployments |\n| `deploy_application` | Step-by-step deployment workflow |\n| `security_audit` | Security scanning and RBAC analysis workflow |\n| `cost_optimization` | Resource optimization and cost analysis workflow |\n| `disaster_recovery` | Backup and recovery planning workflow |\n| `debug_networking` | Network debugging for services and connectivity |\n| `scale_application` | Scaling guide with HPA\u002FVPA best practices |\n| `upgrade_cluster` | Kubernetes cluster upgrade planning |\n\n### Key Capabilities\n\n- 🤖 **253 Powerful Tools** - Complete Kubernetes management from pods to security\n- 🎯 **8 AI Workflow Prompts** - Pre-built workflows for common operations\n- 📊 **8 MCP Resources** - Browsable Kubernetes data exposure\n- 🎨 **6 Interactive Dashboards** - HTML UI tools for visual cluster management\n- 🌐 **26 Browser Tools** - Web automation with cloud provider support\n- 🔄 **107 Ecosystem Tools** - GitOps, Cert-Manager, Policy, Backup, KEDA, Cilium, Rollouts, CAPI, KubeVirt, Istio, vCluster\n- ⚡ **Multi-Transport** - stdio, SSE, HTTP, streamable-http\n- 🔐 **Security First** - Non-destructive mode, secret masking, RBAC validation\n- 🏥 **Advanced Diagnostics** - AI-powered troubleshooting and cost optimization\n- ☸️ **Multi-Cluster** - Target any cluster via context parameter in every tool\n- 🎡 **Full Helm v3** - Complete chart lifecycle management\n- 🔧 **Powerful CLI** - Shell-friendly tool discovery and direct calling\n- 🐳 **Cloud Native** - Deploy in-cluster with kMCP or kagent\n\n## Using the CLI\n\nThe built-in CLI lets you explore and test tools without an AI assistant:\n\n```bash\n# List all tools with descriptions\nkubectl-mcp-server tools -d\n\n# Search for pod-related tools\nkubectl-mcp-server grep \"*pod*\"\n\n# Show specific tool schema\nkubectl-mcp-server tools get_pods\n\n# Call a tool directly\nkubectl-mcp-server call get_pods '{\"namespace\": \"kube-system\"}'\n\n# Pipe JSON from stdin\necho '{\"namespace\": \"default\"}' | kubectl-mcp-server call get_pods\n\n# Check dependencies\nkubectl-mcp-server doctor\n\n# Show\u002Fswitch Kubernetes context\nkubectl-mcp-server context\nkubectl-mcp-server context minikube\n\n# List resources and prompts\nkubectl-mcp-server resources\nkubectl-mcp-server prompts\n\n# Show server info\nkubectl-mcp-server info\n```\n\n### CLI Features\n\n- **Structured errors**: Actionable error messages with suggestions\n- **Colorized output**: Human-readable with JSON mode for scripting (`--json`)\n- **NO_COLOR support**: Respects `NO_COLOR` environment variable\n- **Stdin support**: Pipe JSON arguments to commands\n\n## Advanced Configuration\n\n### Transport Modes\n\nThe server supports multiple transport protocols:\n\n```bash\n# stdio (default) - Best for Claude Desktop, Cursor, Windsurf\nkubectl-mcp-server\n# or: python -m kubectl_mcp_tool.mcp_server\n\n# SSE - Server-Sent Events for web clients\nkubectl-mcp-server --transport sse --port 8000\n\n# HTTP - Standard HTTP for REST clients\nkubectl-mcp-server --transport http --port 8000\n\n# streamable-http - For agentgateway integration\nkubectl-mcp-server --transport streamable-http --port 8000\n```\n\n**Transport Options:**\n- `--transport`: Choose from `stdio`, `sse`, `http`, `streamable-http` (default: `stdio`)\n- `--host`: Bind address (default: `0.0.0.0`)\n- `--port`: Port for network transports (default: `8000`)\n- `--disable-destructive` (or `--non-destructive`): Block destructive operations (allow create\u002Fupdate, block delete)\n- `--read-only`: Block all write operations\n\n### Environment Variables\n\n**Core Settings:**\n\n| Variable | Description | Default |\n|----------|-------------|---------|\n| `KUBECONFIG` | Path to kubeconfig file | `~\u002F.kube\u002Fconfig` |\n| `MCP_DEBUG` | Enable verbose logging | `false` |\n| `MCP_LOG_FILE` | Log file path | None (stdout) |\n\n**Authentication (Enterprise):**\n\n| Variable | Description | Default |\n|----------|-------------|---------|\n| `MCP_AUTH_ENABLED` | Enable OAuth 2.1 authentication | `false` |\n| `MCP_AUTH_ISSUER` | OAuth 2.0 Authorization Server URL | - |\n| `MCP_AUTH_JWKS_URI` | JWKS endpoint URL | Auto-derived |\n| `MCP_AUTH_AUDIENCE` | Expected token audience | `kubectl-mcp-server` |\n| `MCP_AUTH_REQUIRED_SCOPES` | Required OAuth scopes | `mcp:tools` |\n\n**Browser Automation (Optional):**\n\n| Variable | Description | Default |\n|----------|-------------|---------|\n| `MCP_BROWSER_ENABLED` | Enable browser automation tools | `false` |\n| `MCP_BROWSER_PROVIDER` | Cloud provider (browserbase\u002Fbrowseruse) | None |\n| `MCP_BROWSER_PROFILE` | Persistent profile path | None |\n| `MCP_BROWSER_CDP_URL` | Remote CDP WebSocket URL | None |\n| `MCP_BROWSER_PROXY` | Proxy server URL | None |\n\n## Optional: Interactive Dashboards (6 UI Tools)\n\nGet beautiful HTML dashboards for visual cluster management.\n\n**Installation:**\n\n```bash\n# Install with UI support\npip install kubectl-mcp-server[ui]\n```\n\n**6 Dashboard Tools:**\n- 📊 `show_pods_dashboard_ui` - Real-time pod status table\n- 📝 `show_pod_logs_ui` - Interactive log viewer with search\n- 🎯 `show_cluster_overview_ui` - Complete cluster dashboard\n- ⚡ `show_events_timeline_ui` - Events timeline with filtering\n- 📄 `show_resource_yaml_ui` - YAML viewer with syntax highlighting\n- 📸 `render_k8s_dashboard_screenshot` - Export dashboards as PNG\n\n**Features:**\n- 🎨 Dark theme optimized for terminals (Catppuccin)\n- 🔄 Graceful fallback to JSON for incompatible clients\n- 🖼️ Screenshot rendering for universal compatibility\n- 🚀 Zero external dependencies\n\n**Works With**: Goose, LibreChat, Nanobot (full HTML UI) | Claude Desktop, Cursor, others (JSON + screenshots)\n\n## Optional: Browser Automation (26 Tools)\n\nAutomate web-based Kubernetes operations with [agent-browser](https:\u002F\u002Fgithub.com\u002Fvercel-labs\u002Fagent-browser) integration.\n\n**Quick Setup:**\n\n```bash\n# Install agent-browser\nnpm install -g agent-browser\nagent-browser install\n\n# Enable browser tools\nexport MCP_BROWSER_ENABLED=true\nkubectl-mcp-server\n```\n\n**What You Can Do:**\n- 🌐 Test deployed apps via Ingress URLs\n- 📸 Screenshot Grafana, ArgoCD, or any K8s dashboard\n- ☁️ Automate cloud console operations (EKS, GKE, AKS)\n- 🏥 Health check web applications\n- 📄 Export monitoring dashboards as PDF\n- 🔐 Test authentication flows with persistent sessions\n\n**26 Available Tools**: `browser_open`, `browser_screenshot`, `browser_click`, `browser_fill`, `browser_test_ingress`, `browser_screenshot_grafana`, `browser_health_check`, and [19 more](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server#browser-tools)\n\n**Advanced Features**:\n- Cloud providers: Browserbase, Browser Use\n- Persistent browser profiles\n- Remote CDP connections\n- Session management\n\n## Optional: kubectl-mcp-app (8 Interactive UI Dashboards)\n\nA standalone npm package that provides beautiful, interactive UI dashboards for Kubernetes management using the MCP ext-apps SDK.\n\n**Installation:**\n\n```bash\n# Via npm\nnpm install -g kubectl-mcp-app\n\n# Or via npx (no install)\nnpx kubectl-mcp-app\n```\n\n**Claude Desktop Configuration:**\n\n```json\n{\n  \"mcpServers\": {\n    \"kubectl-app\": {\n      \"command\": \"npx\",\n      \"args\": [\"kubectl-mcp-app\"]\n    }\n  }\n}\n```\n\n**8 Interactive UI Tools:**\n\n| Tool | Description |\n| ---- | ----------- |\n| `k8s-pods` | Interactive pod viewer with filtering, sorting, status indicators |\n| `k8s-logs` | Real-time log viewer with syntax highlighting and search |\n| `k8s-deploy` | Deployment dashboard with rollout status, scaling, rollback |\n| `k8s-helm` | Helm release manager with upgrade\u002Frollback actions |\n| `k8s-cluster` | Cluster overview with node health and resource metrics |\n| `k8s-cost` | Cost analyzer with waste detection and recommendations |\n| `k8s-events` | Events timeline with type filtering and grouping |\n| `k8s-network` | Network topology graph showing Services\u002FPods\u002FIngress |\n\n**Features:**\n- 🎨 Dark\u002Flight theme support\n- 📊 Real-time data visualization\n- 🖱️ Interactive actions (scale, restart, delete)\n- 🔗 Seamless integration with kubectl-mcp-server\n\n**More Info**: See [kubectl-mcp-app\u002FREADME.md](.\u002Fkubectl-mcp-app\u002FREADME.md) for full documentation.\n\n## Enterprise: OAuth 2.1 Authentication\n\nSecure your MCP server with OAuth 2.1 authentication (RFC 9728).\n\n```bash\nexport MCP_AUTH_ENABLED=true\nexport MCP_AUTH_ISSUER=https:\u002F\u002Fyour-idp.example.com\nexport MCP_AUTH_AUDIENCE=kubectl-mcp-server\nkubectl-mcp-server --transport http --port 8000\n```\n\n**Supported Identity Providers**: Okta, Auth0, Keycloak, Microsoft Entra ID, Google OAuth, and any OIDC-compliant provider.\n\n**Use Case**: Multi-tenant environments, compliance requirements, audit logging.\n\n## Integrations & Ecosystem\n\n### Docker MCP Toolkit\n\nWorks with [Docker MCP Toolkit](https:\u002F\u002Fdocs.docker.com\u002Fai\u002Fmcp-catalog-and-toolkit\u002Ftoolkit\u002F):\n\n```bash\ndocker mcp server add kubectl-mcp-server mcp\u002Fkubectl-mcp-server:latest\ndocker mcp server configure kubectl-mcp-server --volume \"$HOME\u002F.kube:\u002Froot\u002F.kube:ro\"\ndocker mcp server enable kubectl-mcp-server\ndocker mcp client connect claude\n```\n\n### agentregistry\n\nInstall from the centralized [agentregistry](https:\u002F\u002Faregistry.ai):\n\n```bash\n# Install arctl CLI\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fagentregistry-dev\u002Fagentregistry\u002Fmain\u002Fscripts\u002Finstall.sh | bash\n\n# Install kubectl-mcp-server\narctl mcp install io.github.rohitg00\u002Fkubectl-mcp-server\n```\n\n**Available via**: PyPI (`uvx`), npm (`npx`), OCI (`docker.io\u002Frohitghumare64\u002Fkubectl-mcp-server`)\n\n### agentgateway\n\nRoute to multiple MCP servers through [agentgateway](https:\u002F\u002Fgithub.com\u002Fagentgateway\u002Fagentgateway):\n\n```bash\n# Start with streamable-http\nkubectl-mcp-server --transport streamable-http --port 8000\n\n# Configure gateway\ncat > gateway.yaml \u003C\u003CEOF\nbinds:\n- port: 3000\n  listeners:\n  - routes:\n    - backends:\n      - mcp:\n          targets:\n          - name: kubectl-mcp-server\n            mcp:\n              host: http:\u002F\u002Flocalhost:8000\u002Fmcp\nEOF\n\n# Start gateway\nagentgateway --config gateway.yaml\n```\n\nConnect clients to `http:\u002F\u002Flocalhost:3000\u002Fmcp` for unified access to all 253 tools.\n\n## In-Cluster Deployment\n\n### Option 1: kMCP (Recommended)\n\nDeploy with [kMCP](https:\u002F\u002Fgithub.com\u002Fkagent-dev\u002Fkmcp) - a control plane for MCP servers:\n\n```bash\n# Install kMCP\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fkagent-dev\u002Fkmcp\u002Frefs\u002Fheads\u002Fmain\u002Fscripts\u002Fget-kmcp.sh | bash\nkmcp install\n\n# Deploy kubectl-mcp-server (easiest)\nkmcp deploy package --deployment-name kubectl-mcp-server \\\n   --manager npx --args kubectl-mcp-server\n\n# Or with Docker image\nkmcp deploy --file deploy\u002Fkmcp\u002Fkmcp.yaml --image rohitghumare64\u002Fkubectl-mcp-server:latest\n```\n\nSee [kMCP quickstart](https:\u002F\u002Fkagent.dev\u002Fdocs\u002Fkmcp\u002Fquickstart) for details.\n\n### Option 2: Standard Kubernetes\n\nDeploy with kubectl\u002Fkustomize:\n\n```bash\n# Using kustomize (recommended)\nkubectl apply -k deploy\u002Fkubernetes\u002F\n\n# Or individual manifests\nkubectl apply -f deploy\u002Fkubernetes\u002Fnamespace.yaml\nkubectl apply -f deploy\u002Fkubernetes\u002Frbac.yaml\nkubectl apply -f deploy\u002Fkubernetes\u002Fdeployment.yaml\nkubectl apply -f deploy\u002Fkubernetes\u002Fservice.yaml\n\n# Access via port-forward\nkubectl port-forward -n kubectl-mcp svc\u002Fkubectl-mcp-server 8000:8000\n```\n\nSee [deploy\u002F](deploy\u002F) directory for all manifests and configuration options.\n\n### Option 3: kagent (AI Agent Framework)\n\nIntegrate with [kagent](https:\u002F\u002Fgithub.com\u002Fkagent-dev\u002Fkagent) - a CNCF Kubernetes-native AI agent framework:\n\n```bash\n# Install kagent\nbrew install kagent\nkagent install --profile demo\n\n# Register as ToolServer\nkubectl apply -f deploy\u002Fkagent\u002Ftoolserver-stdio.yaml\n\n# Open dashboard\nkagent dashboard\n```\n\nYour AI agents now have access to all 253 Kubernetes tools. See [kagent quickstart](https:\u002F\u002Fkagent.dev\u002Fdocs\u002Fkagent\u002Fgetting-started\u002Fquickstart).\n\n## Architecture\n\n```\n┌─────────────────┐     ┌──────────────────┐     ┌─────────────────┐\n│   AI Assistant  │────▶│   MCP Server     │────▶│  Kubernetes API │\n│ (Claude\u002FCursor) │◀────│ (kubectl-mcp)    │◀────│    (kubectl)    │\n└─────────────────┘     └──────────────────┘     └─────────────────┘\n```\n\nThe MCP server implements the [Model Context Protocol](https:\u002F\u002Fgithub.com\u002Fmodelcontextprotocol\u002Fspec), translating natural language requests into kubectl operations.\n\n### Modular Structure\n\n```\nkubectl_mcp_tool\u002F\n├── mcp_server.py          # Main server (FastMCP, transports)\n├── tools\u002F                  # 253 MCP tools organized by category\n│   ├── pods.py            # Pod management & diagnostics\n│   ├── deployments.py     # Deployments, StatefulSets, DaemonSets\n│   ├── core.py            # Namespaces, ConfigMaps, Secrets\n│   ├── cluster.py         # Context\u002Fcluster management\n│   ├── networking.py      # Services, Ingress, NetworkPolicies\n│   ├── storage.py         # PVCs, StorageClasses, PVs\n│   ├── security.py        # RBAC, ServiceAccounts, PodSecurity\n│   ├── helm.py            # Complete Helm v3 operations\n│   ├── operations.py      # kubectl apply\u002Fpatch\u002Fdescribe\u002Fetc\n│   ├── diagnostics.py     # Metrics, namespace comparison\n│   ├── cost.py            # Resource optimization & cost analysis\n│   ├── ui.py              # MCP-UI interactive dashboards\n│   ├── gitops.py          # GitOps (Flux\u002FArgoCD)\n│   ├── certs.py           # Cert-Manager\n│   ├── policy.py          # Policy (Kyverno\u002FGatekeeper)\n│   ├── backup.py          # Backup (Velero)\n│   ├── keda.py            # KEDA autoscaling\n│   ├── cilium.py          # Cilium\u002FHubble network observability\n│   ├── rollouts.py        # Argo Rollouts\u002FFlagger\n│   ├── capi.py            # Cluster API\n│   ├── kubevirt.py        # KubeVirt VMs\n│   ├── kiali.py           # Istio\u002FKiali service mesh\n│   └── vind.py            # vCluster (virtual clusters)\n├── resources\u002F              # 8 MCP Resources for data exposure\n├── prompts\u002F                # 8 MCP Prompts for workflows\n└── cli\u002F                    # CLI interface\n```\n\n## Agent Skills (25 Skills for AI Coding Agents)\n\nExtend your AI coding agent with Kubernetes expertise using our [Agent Skills](https:\u002F\u002Fagenstskills.com) library. Skills provide specialized knowledge and workflows that agents can load on demand.\n\n### Quick Install\n\n```bash\n# Copy all skills to Claude\ncp -r kubernetes-skills\u002Fclaude\u002F* ~\u002F.claude\u002Fskills\u002F\n\n# Or install specific skills\ncp -r kubernetes-skills\u002Fclaude\u002Fk8s-helm ~\u002F.claude\u002Fskills\u002F\n```\n\n### Available Skills (25)\n\n| Category | Skills |\n|----------|--------|\n| **Core Resources** | k8s-core, k8s-networking, k8s-storage |\n| **Workloads** | k8s-deploy, k8s-operations, k8s-helm |\n| **Observability** | k8s-diagnostics, k8s-troubleshoot, k8s-incident |\n| **Security** | k8s-security, k8s-policy, k8s-certs |\n| **GitOps** | k8s-gitops, k8s-rollouts |\n| **Scaling** | k8s-autoscaling, k8s-cost, k8s-backup |\n| **Multi-Cluster** | k8s-multicluster, k8s-capi, k8s-kubevirt, k8s-vind |\n| **Networking** | k8s-service-mesh, k8s-cilium |\n| **Tools** | k8s-browser, k8s-cli |\n\n### Convert to Other Agents\n\nUse [SkillKit](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fskillkit) to convert skills to your preferred AI agent format:\n\n```bash\nnpm install -g skillkit\n\n# Convert to Cursor format\nskillkit translate kubernetes-skills\u002Fclaude --to cursor --output .cursor\u002Frules\u002F\n\n# Convert to Codex format\nskillkit translate kubernetes-skills\u002Fclaude --to codex --output .\u002F\n```\n\n**Supported agents:** Claude, Cursor, Codex, Gemini CLI, GitHub Copilot, Goose, Windsurf, Roo, Amp, and more.\n\nSee [kubernetes-skills\u002FREADME.md](kubernetes-skills\u002FREADME.md) for full documentation.\n\n## Multi-Cluster Support\n\nSeamlessly manage multiple Kubernetes clusters through natural language. **Every tool** supports an optional `context` parameter to target any cluster without switching contexts.\n\n### Context Parameter (v1.15.0)\n\nMost kubectl-backed tools accept an optional `context` parameter to target specific clusters.\nNote: vCluster (vind) and kind tools run via their local CLIs and do not accept the `context` parameter.\n\n**Talk to your AI assistant:**\n```\n\"List pods in the production cluster\"\n\"Get deployments from staging context\"\n\"Show logs from the api-pod in the dev cluster\"\n\"Compare namespaces between production and staging clusters\"\n```\n\n**Direct tool calls with context:**\n```bash\n# Target a specific cluster context\nkubectl-mcp-server call get_pods '{\"namespace\": \"default\", \"context\": \"production\"}'\n\n# Get deployments from staging\nkubectl-mcp-server call get_deployments '{\"namespace\": \"app\", \"context\": \"staging\"}'\n\n# Install Helm chart to production cluster\nkubectl-mcp-server call install_helm_chart '{\"name\": \"redis\", \"chart\": \"bitnami\u002Fredis\", \"namespace\": \"cache\", \"context\": \"production\"}'\n\n# Compare resources across clusters\nkubectl-mcp-server call compare_namespaces '{\"namespace1\": \"prod-ns\", \"namespace2\": \"staging-ns\", \"context\": \"production\"}'\n```\n\n### Context Management\n\n**Talk to your AI assistant:**\n```\n\"List all available Kubernetes contexts\"\n\"Switch to the production cluster\"\n\"Show me details about the staging context\"\n\"What's the current cluster I'm connected to?\"\n```\n\n**Or use the CLI directly:**\n```bash\nkubectl-mcp-server context                    # Show current context\nkubectl-mcp-server context production         # Switch context\nkubectl-mcp-server call list_contexts_tool    # List all contexts via MCP\n```\n\n### How It Works\n\n- If `context` is omitted, the tool uses your current kubectl context\n- If `context` is specified, the tool targets that cluster directly\n- Response includes `\"context\": \"production\"` or `\"context\": \"current\"` for clarity\n- Works with all kubeconfig setups and respects `KUBECONFIG` environment variable\n- No need to switch contexts for cross-cluster operations\n\n## Development & Testing\n\n### Setup Development Environment\n\n```bash\n# Clone the repository\ngit clone https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server.git\ncd kubectl-mcp-server\n\n# Create virtual environment\npython -m venv venv\nsource venv\u002Fbin\u002Factivate  # On Windows: venv\\Scripts\\activate\n\n# Install development dependencies\npip install -r requirements-dev.txt\n```\n\n### Running Tests\n\n```bash\n# Run all tests\npytest tests\u002F -v\n\n# Run specific test file\npytest tests\u002Ftest_tools.py -v\n\n# Run with coverage\npytest tests\u002F --cov=kubectl_mcp_tool --cov-report=html\n\n# Run only unit tests\npytest tests\u002F -v -m unit\n```\n\n### Test Structure\n\n```\ntests\u002F\n├── __init__.py          # Test package\n├── conftest.py          # Shared fixtures and mocks\n├── test_tools.py        # Unit tests for 253 MCP tools\n├── test_resources.py    # Tests for 8 MCP Resources\n├── test_prompts.py      # Tests for 8 MCP Prompts\n└── test_server.py       # Server initialization tests\n```\n\n**234 tests covering**: tool registration, resource exposure, prompt generation, server initialization, non-destructive mode, secret masking, error handling, transport methods, CLI commands, browser automation, and ecosystem tools.\n\n### Code Quality\n\n```bash\n# Format code\nblack kubectl_mcp_tool tests\n\n# Sort imports\nisort kubectl_mcp_tool tests\n\n# Lint\nflake8 kubectl_mcp_tool tests\n\n# Type checking\nmypy kubectl_mcp_tool\n```\n\n## Contributing\n\nWe ❤️ contributions! Whether it's bug reports, feature requests, documentation improvements, or code contributions.\n\n**Ways to contribute:**\n- 🐛 Report bugs via [GitHub Issues](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues)\n- 💡 Suggest features or improvements\n- 📝 Improve documentation\n- 🔧 Submit pull requests\n- ⭐ Star the project if you find it useful!\n\n**Development setup**: See [Development & Testing](#development--testing) section above.\n\n**Before submitting a PR:**\n1. Run tests: `pytest tests\u002F -v`\n2. Format code: `black kubectl_mcp_tool tests`\n3. Check linting: `flake8 kubectl_mcp_tool tests`\n\n## Support & Community\n\n- 📖 [Documentation](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server#readme)\n- 💬 [GitHub Discussions](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fdiscussions)\n- 🐛 [Issue Tracker](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues)\n- 🎯 [Feature Requests](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues\u002Fnew)\n- 🌟 [agentregistry Profile](https:\u002F\u002Faregistry.ai)\n\n## License\n\nMIT License - see [LICENSE](LICENSE) for details.\n\n## Links & Resources\n\n**Package Repositories:**\n- 🐍 [PyPI Package](https:\u002F\u002Fpypi.org\u002Fproject\u002Fkubectl-mcp-server\u002F)\n- 📦 [npm Package](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fkubectl-mcp-server)\n- 🐳 [Docker Hub](https:\u002F\u002Fhub.docker.com\u002Fr\u002Frohitghumare64\u002Fkubectl-mcp-server)\n\n**Project:**\n- 🔧 [GitHub Repository](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server)\n- 🐛 [Issue Tracker](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues)\n- 📋 [Changelog](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases)\n\n**Ecosystem:**\n- 📚 [Model Context Protocol](https:\u002F\u002Fmodelcontextprotocol.io)\n- ☸️ [Kubernetes Documentation](https:\u002F\u002Fkubernetes.io\u002Fdocs)\n\n---\n\n**Made with ❤️ for the Kubernetes and AI community**\n\nIf **kubectl-mcp-server** makes your DevOps life easier, give it a ⭐ on [GitHub](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server)!\n","\u003Cp align=\"center\">\n  \u003Cimg src=\"logos\u002Fkubectl-mcp-server-icon.svg\" alt=\"kubectl-mcp-server logo\" width=\"80\" height=\"80\">\n  \u003Cbr>\n  \u003Cstrong style=\"font-size: 24px;\">kubectl-mcp-server\u003C\u002Fstrong>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n\u003Cb>通过与 AI 的自然语言对话，掌控您的整个 Kubernetes 基础设施。\u003C\u002Fb>\u003Cbr>\n像与 DevOps 专家交谈一样与您的集群互动。无需复杂操作，即可调试崩溃的 Pod、优化成本、部署应用、审计安全、管理 Helm 图表以及可视化仪表板——所有这些都可通过自然语言完成。\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002Frohitg00\u002Fkubectl-mcp-server?style=flat&logo=github\" alt=\"GitHub 星标\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fopensource.org\u002Flicenses\u002FMIT\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg\" alt=\"许可证：MIT\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.python.org\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.9+-blue.svg\" alt=\"Python\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fkubernetes.io\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fkubernetes-%23326ce5.svg?style=flat&logo=kubernetes&logoColor=white\" alt=\"Kubernetes\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fmodelcontextprotocol.io\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FMCP-compatible-green.svg\" alt=\"MCP\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fpypi.org\u002Fproject\u002Fkubectl-mcp-server\u002F\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fv\u002Fkubectl-mcp-server?color=blue&label=PyPI\" alt=\"PyPI\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fnpm\u002Fv\u002Fkubectl-mcp-server?color=green&label=npm\" alt=\"npm\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fhub.docker.com\u002Fr\u002Frohitghumare64\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fdocker\u002Fpulls\u002Frohitghumare64\u002Fkubectl-mcp-server.svg\" alt=\"Docker\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Ftests-234%20passed-success\" \n  \u003Ca href=\"https:\u002F\u002Fdeepwiki.com\u002Frohitg00\u002Fkubectl-mcp-server\">\u003Cimg src=\"https:\u002F\u002Fdeepwiki.com\u002Fbadge.svg\" alt=\"Ask DeepWiki\">\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Faregistry.ai\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fagentregistry-verified-blue?logo=data:image\u002Fsvg+xml;base64,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\" alt=\"agentregistry\">\u003C\u002Fa>\n\u003C\u002Fp>\n\n---\n\n## 安装\n\n### 使用 npx 快速入门（推荐——零安装）\n\n```bash\n# 无需安装即可直接运行，即刻生效！\nnpx -y kubectl-mcp-server\n\n# 或者全局安装以加快启动速度\nnpm install -g kubectl-mcp-server\n```\n\n### 或使用 pip 安装（Python）\n\n```bash\n# 标准安装\npip install kubectl-mcp-server\n\n# 推荐安装带交互式 UI 仪表盘的版本\npip install kubectl-mcp-server[ui]\n```\n---\n\n## 📑 目录\n\n- [你能做什么？](#what-can-you-do)\n- [为什么选择 kubectl-mcp-server？](#why-kubectl-mcp-server)\n- [现场演示](#live-demos)\n- [安装](#installation)\n  - [使用 npx 快速入门（推荐——零安装）](#quick-start-with-npx-recommended---zero-install)\n  - [使用 pip 安装（Python）](#or-install-with-pip-python)\n  - [Docker](#docker)\n- [开始使用](#getting-started)\n- [与你的 AI 助手快速设置](#quick-setup-with-your-ai-assistant)\n- [所有支持的 AI 助手](#all-supported-ai-assistants)\n- [完整功能集](#complete-feature-set)\n- [使用 CLI](#using-the-cli)\n- [高级配置](#advanced-configuration)\n- [可选功能](#optional-interactive-dashboards-6-ui-tools)\n  - [交互式仪表盘](#optional-interactive-dashboards-6-ui-tools)\n  - [浏览器自动化](#optional-browser-automation-26-tools)\n- [企业级](#enterprise-oauth-21-authentication)\n- [集成与生态系统](#integrations--ecosystem)\n- [集群内部署](#in-cluster-deployment)\n- [多集群支持](#multi-cluster-support)\n- [架构](#architecture)\n- [代理技能](#agent-skills-24-skills-for-ai-coding-agents)\n- [开发与测试](#development--testing)\n- [贡献代码](#contributing)\n- [支持与社区](#support--community)\n\n---\n\n## 你能做什么？\n\n只需用自然语言向你的 AI 助手提问：\n\n💬 **“我的 Pod 为什么崩溃？”**\n- 即时诊断崩溃原因，提供日志、事件和资源分析\n- 找出根本原因并给出可操作的建议\n\n💬 **“部署一个包含 3 个副本的 Redis 集群”**\n- 按照最佳实践创建部署\n- 自动配置服务、持久化存储和健康检查\n\n💬 **“告诉我哪些 Pod 正在浪费资源？”**\n- 基于 AI 的成本优化分析\n- 提供节省潜力及资源优化建议\n\n💬 **“哪些服务无法访问数据库？”**\n- 网络连通性诊断，包括 DNS 解析\n- 从入口到 Pod 的服务链追踪\n\n💬 **“对所有命名空间进行安全审计”**\n- RBAC 权限分析\n- Secret 安全扫描和 Pod 安全策略检查\n\n💬 **“展示集群仪表板”**\n- 交互式 HTML 仪表盘，实时显示指标\n- 可视化事件和资源使用的时间线\n\n**253 种强大工具** | **8 个工作流提示** | **8 种数据资源** | **兼容所有主流 AI 助手**\n\n## 为什么选择 kubectl-mcp-server？\n\n- **🚀 停止上下文切换** —— 直接从 AI 助手对话中管理 Kubernetes\n- **🧠 AI 驱动的诊断** —— 提供智能故障排除，而非单纯的数据罗列\n- **💰 内置成本优化** —— 发现资源浪费并给出可操作的节约建议\n- **🔒 企业级安全** —— 支持 OAuth 2.1 认证、RBAC 验证、非破坏性模式和 Secret 掩码\n- **⚡ 无学习曲线** —— 使用自然语言代替记忆复杂的 kubectl 命令\n- **🌐 通用兼容性** —— 适用于 Claude、Cursor、Windsurf、Copilot 等 15+ 种 AI 工具\n- **📊 可视化洞察** —— 提供交互式仪表盘和针对 Web 工具的浏览器自动化功能\n- **☸️ 生产级可靠性** —— 可在集群内部署，经过 216 次测试验证，持续维护更新\n\n无论是调试崩溃的 Pod 还是优化集群成本，kubectl-mcp-server 都是您值得信赖的 AI 驱动的 DevOps 伙伴。\n\n## 现场演示\n\n### Claude Desktop\n![Claude MCP](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Frohitg00_kubectl-mcp-server_readme_a1a41d910da7.gif)\n\n### Cursor AI\n![Cursor MCP](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Frohitg00_kubectl-mcp-server_readme_edee1ddf3850.gif)\n\n### Windsurf\n![Windsurf MCP](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Frohitg00_kubectl-mcp-server_readme_d7517ef4c25f.gif)\n\n## 安装\n\n### 使用 npx 快速入门（推荐——零安装）\n\n```bash\n# 无需安装即可直接运行，即刻生效！\nnpx -y kubectl-mcp-server\n\n# 或者全局安装以加快启动速度\nnpm install -g kubectl-mcp-server\n```\n\n### 或使用 pip 安装（Python）\n\n```bash\n# 标准安装\npip install kubectl-mcp-server\n\n# 推荐安装带交互式 UI 仪表盘的版本\npip install kubectl-mcp-server[ui]\n```\n\n### 从 GitHub 发布版本安装\n\n```bash\n# 直接从 GitHub 发布版本安装特定版本（将 {VERSION} 替换为所需版本）\npip install https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases\u002Fdownload\u002Fv{VERSION}\u002Fkubectl_mcp_server-{VERSION}-py3-none-any.whl\n\n# 示例：安装 v1.19.0\npip install https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases\u002Fdownload\u002Fv1.19.0\u002Fkubectl_mcp_server-1.19.0-py3-none-any.whl\n\n# 或者从 Git 仓库安装最新版本\npip install git+https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server.git\n```\n\n### 先决条件\n- **Python 3.9+**（用于 pip 安装）\n- **Node.js 14+**（用于 npx 安装）\n- 已安装并配置好 `kubectl`\n- 可访问 Kubernetes 集群\n\n### Docker\n\n```bash\n# 从 Docker Hub 拉取镜像\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:latest\n\n# 或从 GitHub Container Registry 拉取镜像\ndocker pull ghcr.io\u002Frohitg00\u002Fkubectl-mcp-server:latest\n\n# 使用 stdio 传输运行\ndocker run -i -v $HOME\u002F.kube:\u002Froot\u002F.kube:ro rohitghumare64\u002Fkubectl-mcp-server:latest\n\n# 使用 HTTP 传输运行\ndocker run -p 8000:8000 -v $HOME\u002F.kube:\u002Froot\u002F.kube:ro rohitghumare64\u002Fkubectl-mcp-server:latest --transport sse\n```\n\n## 开始使用\n\n### 1. 测试服务器（可选）\n\n在与您的 AI 助手集成之前，请先验证安装是否成功：\n\n```bash\n# 检查 kubectl 是否已配置\nkubectl cluster-info\n\n# 直接测试 MCP 服务器\nkubectl-mcp-server info\n\n# 列出所有可用工具\nkubectl-mcp-server tools\n\n# 尝试调用一个工具\nkubectl-mcp-server call get_pods '{\"namespace\": \"kube-system\"}'\n```\n\n### 2. 连接到您的 AI 助手\n\n选择您喜欢的 AI 助手，并添加相应的配置：\n\n## 与您的 AI 助手快速设置\n\n### Claude Desktop\n\n添加到 `~\u002FLibrary\u002FApplication Support\u002FClaude\u002Fclaude_desktop_config.json`：\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### Cursor AI\n\n添加到 `~\u002F.cursor\u002Fmcp.json`：\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### Windsurf\n\n添加到 `~\u002F.config\u002Fwindsurf\u002Fmcp.json`：\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### 使用 Python 代替 npx\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"python\",\n      \"args\": [\"-m\", \"kubectl_mcp_tool.mcp_server\"],\n      \"env\": {\n        \"KUBECONFIG\": \"\u002Fpath\u002Fto\u002F.kube\u002Fconfig\"\n      }\n    }\n  }\n}\n```\n\n**更多集成**：GitHub Copilot、Goose、Gemini CLI、Roo Code，以及 [15+ 其他客户端](#mcp-client-compatibility) —> 请参阅下方的[完整配置指南](#all-supported-ai-assistants)。\n\n### 3. 重启您的 AI 助手\n\n添加配置后，请重启您的 AI 助手（GitHub Copilot、Claude Code、Claude Desktop、Cursor 等），以加载 MCP 服务器。\n\n### 4. 尝试以下命令\n\n开始与您的 AI 助手对话，并尝试以下指令：\n\n**故障排除：**\n```\n“给我展示 kube-system 命名空间中的所有 Pod”\n“为什么 nginx-deployment 的 Pod 会崩溃？”\n“诊断 default 命名空间中的网络连接问题”\n```\n\n**部署：**\n```\n“创建一个包含 3 个副本的 nginx 部署”\n“将我的前端部署扩展到 5 个副本”\n“将 api-server 部署回滚到上一个版本”\n```\n\n**成本与优化：**\n```\n“哪些 Pod 正在使用最多的资源？”\n“告诉我有哪些闲置资源正在浪费资金”\n“分析 production 命名空间中的成本优化机会”\n```\n\n**安全：**\n```\n“审计所有命名空间中的 RBAC 权限”\n“检查是否存在不安全的 Secret 和配置”\n“列出具有特权访问权限运行的 Pod”\n```\n\n**Helm：**\n```\n“列出集群中的所有 Helm 发布”\n“从 Bitnami 图表仓库安装 Redis”\n“显示 my nginx-ingress Helm 发布的值”\n```\n\n**多集群：**\n```\n“列出所有可用的 Kubernetes 上下文”\n“切换到 production 集群上下文”\n“显示集群信息和版本”\n```\n\n## MCP 客户端兼容性\n\n可与**所有兼容 MCP 的 AI 助手**无缝协作：\n\n| 客户端         | 状态     | 客户端         | 状态     |\n|----------------|----------|----------------|----------|\n| Claude Desktop | ✅ 原生   | Claude Code    | ✅ 原生   |\n| Cursor         | ✅ 原生   | Windsurf       | ✅ 原生   |\n| GitHub Copilot | ✅ 原生   | OpenAI Codex   | ✅ 原生   |\n| Gemini CLI     | ✅ 原生   | Goose          | ✅ 原生   |\n| Roo Code       | ✅ 原生   | Kilo Code      | ✅ 原生   |\n| Amp            | ✅ 原生   | Trae           | ✅ 原生   |\n| OpenCode       | ✅ 原生   | Kiro CLI       | ✅ 原生   |\n| Antigravity    | ✅ 原生   | Clawdbot       | ✅ 原生   |\n| Droid (Factory)| ✅ 原生   | 任何 MCP 客户端 | ✅ 兼容   |\n\n## 所有支持的 AI 助手\n\n### Claude Code\n\n添加到 `~\u002F.config\u002Fclaude-code\u002Fmcp.json`：\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### GitHub Copilot（VS Code）\n\n添加到 VS Code 的 `settings.json`：\n\n```json\n{\n  \"mcp\": {\n    \"servers\": {\n      \"kubernetes\": {\n        \"command\": \"npx\",\n        \"args\": [\"-y\", \"kubectl-mcp-server\"]\n      }\n    }\n  }\n}\n```\n\n### Goose\n\n添加到 `~\u002F.config\u002Fgoose\u002Fconfig.yaml`：\n\n```yaml\nextensions:\n  kubernetes:\n    command: npx\n    args:\n      - -y\n      - kubectl-mcp-server\n```\n\n### Gemini CLI\n\n添加到 `~\u002F.gemini\u002Fsettings.json`：\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n### Roo Code \u002F Kilo Code\n\n添加到 `~\u002F.config\u002Froo-code\u002Fmcp.json` 或 `~\u002F.config\u002Fkilo-code\u002Fmcp.json`：\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n## 完整功能集\n\n### 253 个用于全面 Kubernetes 管理的 MCP 工具\n\n| 类别 | 工具 |\n|----------|-------|\n| **Pods** | `get_pods`, `get_logs`, `get_pod_events`, `check_pod_health`, `exec_in_pod`, `cleanup_pods`, `get_pod_conditions`, `get_previous_logs` |\n| **Deployments** | `get_deployments`, `create_deployment`, `scale_deployment`, `kubectl_rollout`, `restart_deployment` |\n| **Workloads** | `get_statefulsets`, `get_daemonsets`, `get_jobs`, `get_replicasets` |\n| **Services & Networking** | `get_services`, `get_ingress`, `get_endpoints`, `diagnose_network_connectivity`, `check_dns_resolution`, `trace_service_chain` |\n| **Storage** | `get_persistent_volumes`, `get_pvcs`, `get_storage_classes` |\n| **Config** | `get_configmaps`, `get_secrets`, `get_resource_quotas`, `get_limit_ranges` |\n| **Cluster** | `get_nodes`, `get_namespaces`, `get_cluster_info`, `get_cluster_version`, `health_check`, `get_node_metrics`, `get_pod_metrics` |\n| **RBAC & Security** | `get_rbac_roles`, `get_cluster_roles`, `get_service_accounts`, `audit_rbac_permissions`, `check_secrets_security`, `get_pod_security_info`, `get_admission_webhooks` |\n| **CRDs** | `get_crds`, `get_priority_classes` |\n| **Helm Releases** | `helm_list`, `helm_status`, `helm_history`, `helm_get_values`, `helm_get_manifest`, `helm_get_notes`, `helm_get_hooks`, `helm_get_all` |\n| **Helm Charts** | `helm_show_chart`, `helm_show_values`, `helm_show_readme`, `helm_show_crds`, `helm_show_all`, `helm_search_repo`, `helm_search_hub` |\n| **Helm Repos** | `helm_repo_list`, `helm_repo_add`, `helm_repo_remove`, `helm_repo_update` |\n| **Helm Operations** | `install_helm_chart`, `upgrade_helm_chart`, `uninstall_helm_chart`, `helm_rollback`, `helm_test`, `helm_template`, `helm_template_apply` |\n| **Helm Development** | `helm_create`, `helm_lint`, `helm_package`, `helm_pull`, `helm_dependency_list`, `helm_dependency_update`, `helm_dependency_build`, `helm_version`, `helm_env` |\n| **Context** | `get_current_context`, `switch_context`, `list_contexts`, `list_kubeconfig_contexts` |\n| **Diagnostics** | `diagnose_pod_crash`, `detect_pending_pods`, `get_evicted_pods`, `compare_namespaces` |\n| **Operations** | `kubectl_apply`, `kubectl_create`, `kubectl_describe`, `kubectl_patch`, `delete_resource`, `kubectl_cp`, `backup_resource`, `label_resource`, `annotate_resource`, `taint_node`, `wait_for_condition` |\n| **Autoscaling** | `get_hpa`, `get_pdb` |\n| **Cost Optimization** | `get_resource_recommendations`, `get_idle_resources`, `get_resource_quotas_usage`, `get_cost_analysis`, `get_overprovisioned_resources`, `get_resource_trends`, `get_namespace_cost_allocation`, `optimize_resource_requests` |\n| **Advanced** | `kubectl_generic`, `kubectl_explain`, `get_api_resources`, `port_forward`, `get_resource_usage`, `node_management` |\n| **UI Dashboards** | `show_pod_logs_ui`, `show_pods_dashboard_ui`, `show_resource_yaml_ui`, `show_cluster_overview_ui`, `show_events_timeline_ui`, `render_k8s_dashboard_screenshot` |\n| **GitOps (Flux\u002FArgo)** | `gitops_apps_list`, `gitops_app_get`, `gitops_app_sync`, `gitops_app_status`, `gitops_sources_list`, `gitops_source_get`, `gitops_detect_engine` |\n| **Cert-Manager** | `certs_list`, `certs_get`, `certs_issuers_list`, `certs_issuer_get`, `certs_renew`, `certs_status_explain`, `certs_challenges_list`, `certs_requests_list`, `certs_detect` |\n| **Policy (Kyverno\u002FGatekeeper)** | `policy_list`, `policy_get`, `policy_violations_list`, `policy_explain_denial`, `policy_audit`, `policy_detect` |\n| **Backup (Velero)** | `backup_list`, `backup_get`, `backup_create`, `backup_delete`, `restore_list`, `restore_create`, `restore_get`, `backup_locations_list`, `backup_schedules_list`, `backup_schedule_create`, `backup_detect` |\n| **KEDA Autoscaling** | `keda_scaledobjects_list`, `keda_scaledobject_get`, `keda_scaledjobs_list`, `keda_triggerauths_list`, `keda_triggerauth_get`, `keda_hpa_list`, `keda_detect` |\n| **Cilium\u002FHubble** | `cilium_policies_list`, `cilium_policy_get`, `cilium_endpoints_list`, `cilium_identities_list`, `cilium_nodes_list`, `cilium_status`, `hubble_flows_query`, `cilium_detect` |\n| **Argo Rollouts\u002FFlagger** | `rollouts_list`, `rollout_get`, `rollout_status`, `rollout_promote`, `rollout_abort`, `rollout_retry`, `rollout_restart`, `analysis_runs_list`, `flagger_canaries_list`, `flagger_canary_get`, `rollouts_detect` |\n| **Cluster API** | `capi_clusters_list`, `capi_cluster_get`, `capi_machines_list`, `capi_machine_get`, `capi_machinedeployments_list`, `capi_machinedeployment_scale`, `capi_machinesets_list`, `capi_machinehealthchecks_list`, `capi_clusterclasses_list`, `capi_cluster_kubeconfig`, `capi_detect` |\n| **KubeVirt VMs** | `kubevirt_vms_list`, `kubevirt_vm_get`, `kubevirt_vmis_list`, `kubevirt_vm_start`, `kubevirt_vm_stop`, `kubevirt_vm_restart`, `kubevirt_vm_pause`, `kubevirt_vm_unpause`, `kubevirt_vm_migrate`, `kubevirt_datasources_list`, `kubevirt_instancetypes_list`, `kubevirt_datavolumes_list`, `kubevirt_detect` |\n| **Istio\u002FKiali** | `istio_virtualservices_list`, `istio_virtualservice_get`, `istio_destinationrules_list`, `istio_gateways_list`, `istio_peerauthentications_list`, `istio_authorizationpolicies_list`, `istio_proxy_status`, `istio_analyze`, `istio_sidecar_status`, `istio_detect` |\n| **vCluster (vind)** | `vind_detect_tool`, `vind_list_clusters_tool`, `vind_status_tool`, `vind_get_kubeconfig_tool`, `vind_logs_tool`, `vind_create_cluster_tool`, `vind_delete_cluster_tool`, `vind_pause_tool`, `vind_resume_tool`, `vind_connect_tool`, `vind_disconnect_tool`, `vind_upgrade_tool`, `vind_describe_tool`, `vind_platform_start_tool` |\n| **kind (K8s in Docker)** | `kind_detect_tool`, `kind_version_tool`, `kind_list_clusters_tool`, `kind_get_nodes_tool`, `kind_get_kubeconfig_tool`, `kind_export_logs_tool`, `kind_cluster_info_tool`, `kind_node_labels_tool`, `kind_create_cluster_tool`, `kind_delete_cluster_tool`, `kind_delete_all_clusters_tool`, `kind_load_image_tool`, `kind_load_image_archive_tool`, `kind_build_node_image_tool`, `kind_set_kubeconfig_tool` |\n\n### MCP 资源\n\n以可浏览资源的形式访问 Kubernetes 数据：\n\n| 资源 URI | 描述 |\n|--------------|-------------|\n| `kubeconfig:\u002F\u002Fcontexts` | 列出所有可用的 kubectl 上下文 |\n| `kubeconfig:\u002F\u002Fcurrent-context` | 获取当前活动上下文 |\n| `namespace:\u002F\u002Fcurrent` | 获取当前命名空间 |\n| `namespace:\u002F\u002Flist` | 列出所有命名空间 |\n| `cluster:\u002F\u002Finfo` | 获取集群信息 |\n| `cluster:\u002F\u002Fnodes` | 获取详细的节点信息 |\n| `cluster:\u002F\u002Fversion` | 获取 Kubernetes 版本 |\n| `cluster:\u002F\u002Fapi-resources` | 列出可用的 API 资源 |\n| `manifest:\u002F\u002Fdeployments\u002F{ns}\u002F{name}` | 获取 Deployment YAML |\n| `manifest:\u002F\u002Fservices\u002F{ns}\u002F{name}` | 获取 Service YAML |\n| `manifest:\u002F\u002Fpods\u002F{ns}\u002F{name}` | 获取 Pod YAML |\n| `manifest:\u002F\u002Fconfigmaps\u002F{ns}\u002F{name}` | 获取 ConfigMap YAML |\n| `manifest:\u002F\u002Fsecrets\u002F{ns}\u002F{name}` | 获取 Secret YAML（数据已屏蔽）|\n| `manifest:\u002F\u002Fingresses\u002F{ns}\u002F{name}` | 获取 Ingress YAML |\n\n### MCP 提示词\n\n针对常见 Kubernetes 操作的预构建工作流提示词：\n\n| 提示词 | 描述 |\n|--------|-------------|\n| `troubleshoot_workload` | 针对 Pod\u002FDeployment 的全面故障排除指南 |\n| `deploy_application` | 分步部署工作流 |\n| `security_audit` | 安全扫描与 RBAC 分析工作流 |\n| `cost_optimization` | 资源优化与成本分析工作流 |\n| `disaster_recovery` | 备份与恢复规划工作流 |\n| `debug_networking` | 服务与连通性的网络调试 |\n| `scale_application` | 带有 HPA\u002FVPA 最佳实践的扩缩容指南 |\n| `upgrade_cluster` | Kubernetes 集群升级规划 |\n\n### 核心能力\n\n- 🤖 **253 个强大工具** - 从 Pod 到安全，实现完整的 Kubernetes 管理\n- 🎯 **8 个 AI 工作流提示词** - 针对常见操作的预构建工作流\n- 📊 **8 个 MCP 资源** - 可浏览的 Kubernetes 数据展示\n- 🎨 **6 个交互式仪表板** - 用于可视化集群管理的 HTML UI 工具\n- 🌐 **26 个浏览器工具** - 支持云提供商的 Web 自动化\n- 🔄 **107 个生态工具** - GitOps、Cert-Manager、Policy、Backup、KEDA、Cilium、Rollouts、CAPI、KubeVirt、Istio、vCluster\n- ⚡ **多传输方式** - stdio、SSE、HTTP、streamable-http\n- 🔐 **安全优先** - 非破坏性模式、密钥掩码、RBAC 验证\n- 🏥 **高级诊断** - 基于 AI 的故障排除和成本优化\n- ☸️ **多集群支持** - 通过每个工具中的上下文参数可靶向任意集群\n- 🎡 **完整 Helm v3** - 全生命周期 Chart 管理\n- 🔧 **强大的 CLI** - 符合 Shell 风格的工具发现与直接调用\n- 🐳 **云原生** - 可在集群内使用 kMCP 或 kagent 部署\n\n## 使用 CLI\n\n内置 CLI 让您无需 AI 助手即可探索和测试工具：\n\n```bash\n# 列出所有工具及其描述\nkubectl-mcp-server tools -d\n\n# 搜索与 Pod 相关的工具\nkubectl-mcp-server grep \"*pod*\"\n\n# 显示特定工具的 Schema\nkubectl-mcp-server tools get_pods\n\n# 直接调用工具\nkubectl-mcp-server call get_pods '{\"namespace\": \"kube-system\"}'\n\n# 从 stdin 管道传递 JSON\necho '{\"namespace\": \"default\"}' | kubectl-mcp-server call get_pods\n\n# 检查依赖项\nkubectl-mcp-server doctor\n\n# 显示或切换 Kubernetes 上下文\nkubectl-mcp-server context\nkubectl-mcp-server context minikube\n\n# 列出资源和提示词\nkubectl-mcp-server resources\nkubectl-mcp-server prompts\n\n# 显示服务器信息\nkubectl-mcp-server info\n```\n\n### CLI 特性\n\n- **结构化错误**：提供可操作的错误消息及建议\n- **彩色输出**：便于人类阅读，并支持 JSON 模式以供脚本使用 (`--json`)\n- **NO_COLOR 支持**：尊重 `NO_COLOR` 环境变量\n- **Stdin 支持**：可将 JSON 参数管道传递给命令\n\n## 高级配置\n\n### 传输模式\n\n服务器支持多种传输协议：\n\n```bash\n# stdio（默认） - 适用于 Claude Desktop、Cursor、Windsurf\nkubectl-mcp-server\n# 或：python -m kubectl_mcp_tool.mcp_server\n\n# SSE - 用于 Web 客户端的服务器发送事件\nkubectl-mcp-server --transport sse --port 8000\n\n# HTTP - 用于 REST 客户端的标准 HTTP\nkubectl-mcp-server --transport http --port 8000\n\n# streamable-http - 用于 agentgateway 集成\nkubectl-mcp-server --transport streamable-http --port 8000\n```\n\n**传输选项：**\n- `--transport`：选择 `stdio`、`sse`、`http`、`streamable-http`（默认：`stdio`）\n- `--host`：绑定地址（默认：`0.0.0.0`）\n- `--port`：网络传输使用的端口（默认：`8000`）\n- `--disable-destructive`（或 `--non-destructive`）：阻止破坏性操作（允许创建\u002F更新，禁止删除）\n- `--read-only`：阻止所有写操作\n\n### 环境变量\n\n**核心设置：**\n\n| 变量 | 描述 | 默认值 |\n|----------|-------------|---------|\n| `KUBECONFIG` | kubeconfig 文件路径 | `~\u002F.kube\u002Fconfig` |\n| `MCP_DEBUG` | 启用详细日志记录 | `false` |\n| `MCP_LOG_FILE` | 日志文件路径 | 无（stdout） |\n\n**认证（企业版）：**\n\n| 变量 | 描述 | 默认值 |\n|----------|-------------|---------|\n| `MCP_AUTH_ENABLED` | 启用 OAuth 2.1 认证 | `false` |\n| `MCP_AUTH_ISSUER` | OAuth 2.0 授权服务器 URL | - |\n| `MCP_AUTH_JWKS_URI` | JWKS 端点 URL | 自动推导 |\n| `MCP_AUTH_AUDIENCE` | 预期的令牌受众 | `kubectl-mcp-server` |\n| `MCP_AUTH_REQUIRED_SCOPES` | 必需的 OAuth 范围 | `mcp:tools` |\n\n**浏览器自动化（可选）：**\n\n| 变量 | 描述 | 默认值 |\n|----------|-------------|---------|\n| `MCP_BROWSER_ENABLED` | 启用浏览器自动化工具 | `false` |\n| `MCP_BROWSER_PROVIDER` | 云提供商（browserbase\u002Fbrowseruse） | 无 |\n| `MCP_BROWSER_PROFILE` | 持久化配置文件路径 | 无 |\n| `MCP_BROWSER_CDP_URL` | 远程 CDP WebSocket URL | 无 |\n| `MCP_BROWSER_PROXY` | 代理服务器 URL | 无 |\n\n## 可选：交互式仪表板（6 个 UI 工具）\n\n获取精美的 HTML 仪表板，用于可视化集群管理。\n\n**安装：**\n\n```bash\n# 安装带 UI 支持的版本\npip install kubectl-mcp-server[ui]\n```\n\n**6 个仪表板工具：**\n- 📊 `show_pods_dashboard_ui` - 实时 Pod 状态表格\n- 📝 `show_pod_logs_ui` - 带搜索功能的交互式日志查看器\n- 🎯 `show_cluster_overview_ui` - 完整的集群仪表板\n- ⚡ `show_events_timeline_ui` - 带过滤功能的事件时间线\n- 📄 `show_resource_yaml_ui` - 带语法高亮的 YAML 查看器\n- 📸 `render_k8s_dashboard_screenshot` - 将仪表板导出为 PNG\n\n**特性：**\n- 🎨 针对终端优化的暗色主题（Catppuccin）\n- 🔄 对不兼容客户端优雅地回退到 JSON\n- 🖼️ 截图渲染以确保通用兼容性\n- 🚀 无外部依赖\n\n**兼容性**：Goose、LibreChat、Nanobot（完整 HTML UI）| Claude Desktop、Cursor 等（JSON + 截图）\n\n## 可选：浏览器自动化（26 个工具）\n\n通过 [agent-browser](https:\u002F\u002Fgithub.com\u002Fvercel-labs\u002Fagent-browser) 集成，自动执行基于 Web 的 Kubernetes 操作。\n\n**快速设置：**\n\n```bash\n# 安装 agent-browser\nnpm install -g agent-browser\nagent-browser install\n\n# 启用浏览器工具\nexport MCP_BROWSER_ENABLED=true\nkubectl-mcp-server\n```\n\n**您可以做的事情：**\n- 🌐 通过 Ingress URL 测试已部署的应用\n- 📸 截取 Grafana、ArgoCD 或任何 K8s 仪表板的截图\n- ☁️ 自动化云控制台操作（EKS、GKE、AKS）\n- 🏥 对 Web 应用进行健康检查\n- 📄 将监控仪表板导出为 PDF\n- 🔐 使用持久化会话测试认证流程\n\n**可用的 26 个工具**：`browser_open`、`browser_screenshot`、`browser_click`、`browser_fill`、`browser_test_ingress`、`browser_screenshot_grafana`、`browser_health_check`，以及 [另外 19 个](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server#browser-tools)\n\n**高级特性：**\n- 云提供商：Browserbase、Browser Use\n- 持久化浏览器配置文件\n- 远程 CDP 连接\n- 会话管理\n\n## 可选：kubectl-mcp-app（8 个交互式 UI 仪表板）\n\n一个独立的 npm 包，利用 MCP ext-apps SDK 提供美观、交互式的 UI 仪表板，用于 Kubernetes 管理。\n\n**安装：**\n\n```bash\n# 安装 kubectl-mcp-app\nnpm install kubectl-mcp-app\n```\n\n# 通过 npm\nnpm install -g kubectl-mcp-app\n\n# 或者通过 npx（无需安装）\nnpx kubectl-mcp-app\n```\n\n**Claude Desktop 配置：**\n\n```json\n{\n  \"mcpServers\": {\n    \"kubectl-app\": {\n      \"command\": \"npx\",\n      \"args\": [\"kubectl-mcp-app\"]\n    }\n  }\n}\n```\n\n**8 个交互式 UI 工具：**\n\n| 工具 | 描述 |\n| ---- | ----------- |\n| `k8s-pods` | 带有筛选、排序和状态指示器的交互式 Pod 查看器 |\n| `k8s-logs` | 支持语法高亮和搜索的实时日志查看器 |\n| `k8s-deploy` | 包含发布状态、扩缩容和回滚功能的部署仪表板 |\n| `k8s-helm` | 提供升级\u002F回滚操作的 Helm 发布管理器 |\n| `k8s-cluster` | 显示节点健康状况和资源指标的集群概览 |\n| `k8s-cost` | 具有浪费检测和建议的成本分析工具 |\n| `k8s-events` | 支持按类型过滤和分组的事件时间线 |\n| `k8s-network` | 展示 Service\u002FPod\u002FIngress 的网络拓扑图 |\n\n**功能：**\n- 🎨 暗色\u002F亮色主题支持\n- 📊 实时数据可视化\n- 🖱️ 交互式操作（扩缩容、重启、删除）\n- 🔗 与 kubectl-mcp-server 的无缝集成\n\n**更多信息**：请参阅 [kubectl-mcp-app\u002FREADME.md](.\u002Fkubectl-mcp-app\u002FREADME.md) 获取完整文档。\n\n## 企业级：OAuth 2.1 认证\n\n使用 OAuth 2.1 认证（RFC 9728）保护您的 MCP 服务器。\n\n```bash\nexport MCP_AUTH_ENABLED=true\nexport MCP_AUTH_ISSUER=https:\u002F\u002Fyour-idp.example.com\nexport MCP_AUTH_AUDIENCE=kubectl-mcp-server\nkubectl-mcp-server --transport http --port 8000\n```\n\n**支持的身份提供商**：Okta、Auth0、Keycloak、Microsoft Entra ID、Google OAuth，以及任何符合 OIDC 标准的提供商。\n\n**适用场景**：多租户环境、合规性要求、审计日志记录。\n\n## 集成与生态系统\n\n### Docker MCP 工具包\n\n可与 [Docker MCP 工具包](https:\u002F\u002Fdocs.docker.com\u002Fai\u002Fmcp-catalog-and-toolkit\u002Ftoolkit\u002F) 一起使用：\n\n```bash\ndocker mcp server add kubectl-mcp-server mcp\u002Fkubectl-mcp-server:latest\ndocker mcp server configure kubectl-mcp-server --volume \"$HOME\u002F.kube:\u002Froot\u002F.kube:ro\"\ndocker mcp server enable kubectl-mcp-server\ndocker mcp client connect claude\n```\n\n### agentregistry\n\n从中心化的 [agentregistry](https:\u002F\u002Faregistry.ai) 安装：\n\n```bash\n# 安装 arctl CLI\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fagentregistry-dev\u002Fagentregistry\u002Fmain\u002Fscripts\u002Finstall.sh | bash\n\n# 安装 kubectl-mcp-server\narctl mcp install io.github.rohitg00\u002Fkubectl-mcp-server\n```\n\n**可通过以下方式获取**：PyPI (`uvx`)、npm (`npx`)、OCI (`docker.io\u002Frohitghumare64\u002Fkubectl-mcp-server`)\n\n### agentgateway\n\n通过 [agentgateway](https:\u002F\u002Fgithub.com\u002Fagentgateway\u002Fagentgateway) 路由到多个 MCP 服务器：\n\n```bash\n# 使用 streamable-http 启动\nkubectl-mcp-server --transport streamable-http --port 8000\n\n# 配置网关\ncat > gateway.yaml \u003C\u003CEOF\nbinds:\n- port: 3000\n  listeners:\n  - routes:\n    - backends:\n      - mcp:\n          targets:\n          - name: kubectl-mcp-server\n            mcp:\n              host: http:\u002F\u002Flocalhost:8000\u002Fmcp\nEOF\n\n# 启动网关\nagentgateway --config gateway.yaml\n```\n\n客户端可连接到 `http:\u002F\u002Flocalhost:3000\u002Fmcp` 来统一访问所有 253 个工具。\n\n## 集群内部署\n\n### 选项 1：kMCP（推荐）\n\n使用 [kMCP](https:\u002F\u002Fgithub.com\u002Fkagent-dev\u002Fkmcp) 部署——一个用于 MCP 服务器的控制平面：\n\n```bash\n# 安装 kMCP\ncurl -fsSL https:\u002F\u002Fraw.githubusercontent.com\u002Fkagent-dev\u002Fkmcp\u002Frefs\u002Fheads\u002Fmain\u002Fscripts\u002Fget-kmcp.sh | bash\nkmcp install\n\n# 部署 kubectl-mcp-server（最简单）\nkmcp deploy package --deployment-name kubectl-mcp-server \\\n   --manager npx --args kubectl-mcp-server\n\n# 或者使用 Docker 镜像\nkmcp deploy --file deploy\u002Fkmcp\u002Fkmcp.yaml --image rohitghumare64\u002Fkubectl-mcp-server:latest\n```\n\n详情请参阅 [kMCP 快速入门](https:\u002F\u002Fkagent.dev\u002Fdocs\u002Fkmcp\u002Fquickstart)。\n\n### 选项 2：标准 Kubernetes\n\n使用 kubectl\u002Fkustomize 部署：\n\n```bash\n# 推荐使用 kustomize\nkubectl apply -k deploy\u002Fkubernetes\u002F\n\n# 或者单独应用清单\nkubectl apply -f deploy\u002Fkubernetes\u002Fnamespace.yaml\nkubectl apply -f deploy\u002Fkubernetes\u002Frbac.yaml\nkubectl apply -f deploy\u002Fkubernetes\u002Fdeployment.yaml\nkubectl apply -f deploy\u002Fkubernetes\u002Fservice.yaml\n\n# 通过端口转发访问\nkubectl port-forward -n kubectl-mcp svc\u002Fkubectl-mcp-server 8000:8000\n```\n\n所有清单和配置选项请参阅 [deploy\u002F](deploy\u002F) 目录。\n\n### 选项 3：kagent（AI 代理框架）\n\n与 [kagent](https:\u002F\u002Fgithub.com\u002Fkagent-dev\u002Fkagent) 集成——一个 CNCF Kubernetes 原生的 AI 代理框架：\n\n```bash\n# 安装 kagent\nbrew install kagent\nkagent install --profile demo\n\n# 注册为 ToolServer\nkubectl apply -f deploy\u002Fkagent\u002Ftoolserver-stdio.yaml\n\n# 打开仪表板\nkagent dashboard\n```\n\n您的 AI 代理现在可以访问所有 253 个 Kubernetes 工具。详情请参阅 [kagent 快速入门](https:\u002F\u002Fkagent.dev\u002Fdocs\u002Fkagent\u002Fgetting-started\u002Fquickstart)。\n\n## 架构\n\n```\n┌─────────────────┐     ┌──────────────────┐     ┌─────────────────┐\n│   AI 助手  │────▶│   MCP 服务器     │────▶│  Kubernetes API │\n│ (Claude\u002FCursor) │◀────│ (kubectl-mcp)    │◀────│    (kubectl)    │\n└─────────────────┘     └──────────────────┘     └─────────────────┘\n```\n\nMCP 服务器实现了 [Model Context Protocol](https:\u002F\u002Fgithub.com\u002Fmodelcontextprotocol\u002Fspec)，将自然语言请求转换为 kubectl 操作。\n\n### 模块化结构\n\n```\nkubectl_mcp_tool\u002F\n├── mcp_server.py          # 主服务器（FastMCP，传输协议）\n├── tools\u002F                  # 按类别组织的 253 个 MCP 工具\n│   ├── pods.py            # Pod 管理与诊断\n│   ├── deployments.py     # Deployments、StatefulSets、DaemonSets\n│   ├── core.py            # Namespaces、ConfigMaps、Secrets\n│   ├── cluster.py         # 上下文\u002F集群管理\n│   ├── networking.py      # Services、Ingress、NetworkPolicies\n│   ├── storage.py         # PVCs、StorageClasses、PVs\n│   ├── security.py        # RBAC、ServiceAccounts、PodSecurity\n│   ├── helm.py            # 完整的 Helm v3 操作\n│   ├── operations.py      # kubectl apply\u002Fpatch\u002Fdescribe 等\n│   ├── diagnostics.py     # 指标、命名空间比较\n│   ├── cost.py            # 资源优化与成本分析\n│   ├── ui.py              # MCP-UI 交互式仪表板\n│   ├── gitops.py          # GitOps（Flux\u002FArgoCD）\n│   ├── certs.py           # Cert-Manager\n│   ├── policy.py          # Policy（Kyverno\u002FGatekeeper）\n│   ├── backup.py          # Backup（Velero）\n│   ├── keda.py            # KEDA 自动缩放\n│   ├── cilium.py          # Cilium\u002FHubble 网络可观ability\n│   ├── rollouts.py        # Argo Rollouts\u002FFlagger\n│   ├── capi.py            # Cluster API\n│   ├── kubevirt.py        # KubeVirt VMs\n│   ├── kiali.py           # Istio\u002FKiali 服务网格\n│   └── vind.py            # vCluster（虚拟集群）\n├── resources\u002F              # 8 个用于数据暴露的 MCP 资源\n├── prompts\u002F                # 8 个用于工作流的 MCP 提示\n└── cli\u002F                    # CLI 界面\n```\n\n## 代理技能（适用于 AI 编码代理的 25 种技能）\n\n借助我们的 [Agent Skills](https:\u002F\u002Fagenstskills.com) 库，为您的 AI 编码代理扩展 Kubernetes 专业知识。这些技能提供了代理可按需加载的专业知识和工作流。\n\n### 快速安装\n\n```bash\n# 将所有技能复制到 Claude\ncp -r kubernetes-skills\u002Fclaude\u002F* ~\u002F.claude\u002Fskills\u002F\n\n# 或者仅安装特定技能\ncp -r kubernetes-skills\u002Fclaude\u002Fk8s-helm ~\u002F.claude\u002Fskills\u002F\n```\n\n### 可用技能（25 种）\n\n| 类别 | 技能 |\n|----------|--------|\n| **核心资源** | k8s-core, k8s-networking, k8s-storage |\n| **工作负载** | k8s-deploy, k8s-operations, k8s-helm |\n| **可观测性** | k8s-diagnostics, k8s-troubleshoot, k8s-incident |\n| **安全** | k8s-security, k8s-policy, k8s-certs |\n| **GitOps** | k8s-gitops, k8s-rollouts |\n| **扩展** | k8s-autoscaling, k8s-cost, k8s-backup |\n| **多集群** | k8s-multicluster, k8s-capi, k8s-kubevirt, k8s-vind |\n| **网络** | k8s-service-mesh, k8s-cilium |\n| **工具** | k8s-browser, k8s-cli |\n\n### 转换为其他代理\n\n使用 [SkillKit](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fskillkit) 将技能转换为您偏好的 AI 代理格式：\n\n```bash\nnpm install -g skillkit\n\n# 转换为 Cursor 格式\nskillkit translate kubernetes-skills\u002Fclaude --to cursor --output .cursor\u002Frules\u002F\n\n# 转换为 Codex 格式\nskillkit translate kubernetes-skills\u002Fclaude --to codex --output .\u002F\n```\n\n**支持的代理：** Claude、Cursor、Codex、Gemini CLI、GitHub Copilot、Goose、Windsurf、Roo、Amp 等。\n\n完整文档请参阅 [kubernetes-skills\u002FREADME.md](kubernetes-skills\u002FREADME.md)。\n\n## 多集群支持\n\n通过自然语言无缝管理多个 Kubernetes 集群。**所有工具**均支持可选的 `context` 参数，以便在不切换上下文的情况下直接操作任意集群。\n\n### 上下文参数（v1.15.0）\n\n大多数基于 kubectl 的工具都接受可选的 `context` 参数，用于指定目标集群。注意：vCluster（vind）和 kind 工具是通过其本地 CLI 运行的，因此不接受 `context` 参数。\n\n**与您的 AI 助手对话：**\n```\n“列出生产集群中的 Pod”\n“从 staging 上下文中获取部署”\n“显示 dev 集群中 api-pod 的日志”\n“比较生产与 staging 集群之间的命名空间”\n```\n\n**直接调用工具并指定上下文：**\n```bash\n# 指定特定集群上下文\nkubectl-mcp-server call get_pods '{\"namespace\": \"default\", \"context\": \"production\"}'\n\n# 从 staging 获取部署\nkubectl-mcp-server call get_deployments '{\"namespace\": \"app\", \"context\": \"staging\"}'\n\n# 在生产集群中安装 Helm Chart\nkubectl-mcp-server call install_helm_chart '{\"name\": \"redis\", \"chart\": \"bitnami\u002Fredis\", \"namespace\": \"cache\", \"context\": \"production\"}'\n\n# 比较跨集群资源\nkubectl-mcp-server call compare_namespaces '{\"namespace1\": \"prod-ns\", \"namespace2\": \"staging-ns\", \"context\": \"production\"}'\n```\n\n### 上下文管理\n\n**与您的 AI 助手对话：**\n```\n“列出所有可用的 Kubernetes 上下文”\n“切换到生产集群”\n“向我展示 staging 上下文的详细信息”\n“我现在连接的是哪个集群？”\n```\n\n**或直接使用 CLI：**\n```bash\nkubectl-mcp-server context                    # 显示当前上下文\nkubectl-mcp-server context production         # 切换上下文\nkubectl-mcp-server call list_contexts_tool    # 通过 MCP 列出所有上下文\n```\n\n### 工作原理\n\n- 如果未指定 `context`，工具将使用您当前的 kubectl 上下文。\n- 如果指定了 `context`，工具将直接操作该集群。\n- 响应中会包含 `\"context\": \"production\"` 或 `\"context\": \"current\"`，以明确上下文。\n- 适用于所有 kubeconfig 配置，并尊重 `KUBECONFIG` 环境变量。\n- 无需切换上下文即可进行跨集群操作。\n\n## 开发与测试\n\n### 设置开发环境\n\n```bash\n# 克隆仓库\ngit clone https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server.git\ncd kubectl-mcp-server\n\n# 创建虚拟环境\npython -m venv venv\nsource venv\u002Fbin\u002Factivate  # Windows 系统：venv\\Scripts\\activate\n\n# 安装开发依赖\npip install -r requirements-dev.txt\n```\n\n### 运行测试\n\n```bash\n# 运行所有测试\npytest tests\u002F -v\n\n# 运行特定测试文件\npytest tests\u002Ftest_tools.py -v\n\n# 带覆盖率运行\npytest tests\u002F --cov=kubectl_mcp_tool --cov-report=html\n\n# 仅运行单元测试\npytest tests\u002F -v -m unit\n```\n\n### 测试结构\n\n```\ntests\u002F\n├── __init__.py          # 测试包\n├── conftest.py          # 共享 fixture 和 mock\n├── test_tools.py        # 253 个 MCP 工具的单元测试\n├── test_resources.py    # 8 个 MCP 资源的测试\n├── test_prompts.py      # 8 个 MCP 提示词的测试\n└── test_server.py       # 服务器初始化测试\n```\n\n**234 个测试涵盖：** 工具注册、资源暴露、提示词生成、服务器初始化、非破坏性模式、敏感信息屏蔽、错误处理、传输方法、CLI 命令、浏览器自动化以及生态系统工具。\n\n### 代码质量\n\n```bash\n# 格式化代码\nblack kubectl_mcp_tool tests\n\n# 排序导入\nisort kubectl_mcp_tool tests\n\n# 静态代码分析\nflake8 kubectl_mcp_tool tests\n\n# 类型检查\nmypy kubectl_mcp_tool\n```\n\n## 贡献\n\n我们非常欢迎各种形式的贡献！无论是 bug 报告、功能请求、文档改进，还是代码贡献。\n\n**贡献方式：**\n- 🐛 通过 [GitHub Issues](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues) 报告 bug\n- 💡 提出功能或改进建议\n- 📝 改进文档\n- 🔧 提交 pull request\n- ⭐ 如果您觉得项目有用，请为它点赞！\n\n**开发设置：** 请参阅上方的 [开发与测试](#development--testing) 部分。\n\n**提交 PR 前：**\n1. 运行测试：`pytest tests\u002F -v`\n2. 格式化代码：`black kubectl_mcp_tool tests`\n3. 检查静态代码：`flake8 kubectl_mcp_tool tests`\n\n## 支持与社区\n\n- 📖 [文档](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server#readme)\n- 💬 [GitHub Discussions](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fdiscussions)\n- 🐛 [问题追踪器](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues)\n- 🎯 [功能请求](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues\u002Fnew)\n- 🌟 [agentregistry 个人主页](https:\u002F\u002Faregistry.ai)\n\n## 许可证\n\nMIT 许可证 - 详情请参阅 [LICENSE](LICENSE) 文件。\n\n## 链接与资源\n\n**软件包仓库：**\n- 🐍 [PyPI 包](https:\u002F\u002Fpypi.org\u002Fproject\u002Fkubectl-mcp-server\u002F)\n- 📦 [npm 包](https:\u002F\u002Fwww.npmjs.com\u002Fpackage\u002Fkubectl-mcp-server)\n- 🐳 [Docker Hub](https:\u002F\u002Fhub.docker.com\u002Fr\u002Frohitghumare64\u002Fkubectl-mcp-server)\n\n**项目：**\n- 🔧 [GitHub 仓库](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server)\n- 🐛 [问题跟踪器](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues)\n- 📋 [变更日志](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases)\n\n**生态系统：**\n- 📚 [模型上下文协议](https:\u002F\u002Fmodelcontextprotocol.io)\n- ☸️ [Kubernetes 官方文档](https:\u002F\u002Fkubernetes.io\u002Fdocs)\n\n---\n\n**专为 Kubernetes 和 AI 社区用心打造**\n\n如果 **kubectl-mcp-server** 让您的 DevOps 工作更加轻松，请在 [GitHub](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server) 上给它点个赞吧！","# kubectl-mcp-server 快速上手指南\n\n**kubectl-mcp-server** 是一个基于 Model Context Protocol (MCP) 的开源工具，让你能够通过自然语言与 AI 助手对话，直接管理和诊断 Kubernetes 集群。无需记忆复杂的 `kubectl` 命令，即可实现故障排查、应用部署、成本优化和安全审计。\n\n## 1. 环境准备\n\n在开始之前，请确保你的开发环境满足以下要求：\n\n*   **操作系统**: Linux, macOS 或 Windows (WSL2 推荐)\n*   **Kubernetes 访问权限**:\n    *   已安装并配置好 `kubectl` 命令行工具。\n    *   拥有有效的 `kubeconfig` 文件（通常位于 `~\u002F.kube\u002Fconfig`），且能正常连接集群。\n    *   验证命令：`kubectl cluster-info`\n*   **运行环境 (二选一)**:\n    *   **Node.js**: 版本 14+ (推荐使用 `npx` 方式，无需手动安装)。\n    *   **Python**: 版本 3.9+ (如果使用 `pip` 安装)。\n*   **AI 助手客户端**: 支持 MCP 协议的客户端，如 Claude Desktop, Cursor, Windsurf, GitHub Copilot 等。\n\n> **国内开发者提示**: 如果下载 npm 或 pip 包速度较慢，建议配置国内镜像源。\n> *   npm: `npm config set registry https:\u002F\u002Fregistry.npmmirror.com`\n> *   pip: `pip config set global.index-url https:\u002F\u002Fpypi.tuna.tsinghua.edu.cn\u002Fsimple`\n\n## 2. 安装步骤\n\n你可以选择通过 `npx` (推荐，零安装) 或 `pip` 进行安装。\n\n### 方案 A：使用 npx (推荐)\n无需全局安装，直接运行即可，最适合快速尝试。\n\n```bash\n# 直接运行 (首次会自动下载)\nnpx -y kubectl-mcp-server\n\n# 或者全局安装以加快启动速度\nnpm install -g kubectl-mcp-server\n```\n\n### 方案 B：使用 pip (Python)\n如果你更习惯 Python 环境，或需要启用交互式 UI 仪表盘功能。\n\n```bash\n# 标准安装\npip install kubectl-mcp-server\n\n# 推荐：安装包含交互式 UI 仪表盘的版本\npip install kubectl-mcp-server[ui]\n```\n\n### 方案 C：Docker 运行\n适合容器化环境或不想污染本地环境的场景。\n\n```bash\n# 拉取镜像并运行 (挂载 kubeconfig)\ndocker run -i -v $HOME\u002F.kube:\u002Froot\u002F.kube:ro rohitghumare64\u002Fkubectl-mcp-server:latest\n```\n\n## 3. 基本使用\n\n安装完成后，你需要将工具配置到你的 AI 助手客户端中，然后通过自然语言发起指令。\n\n### 第一步：配置 AI 助手\n\n根据你的 AI 客户端，将以下配置添加到对应的配置文件中。配置后需**重启 AI 客户端**。\n\n#### 选项 1: Claude Desktop\n编辑文件：`~\u002FLibrary\u002FApplication Support\u002FClaude\u002Fclaude_desktop_config.json` (Mac) 或 `%APPDATA%\\Claude\\claude_desktop_config.json` (Windows)\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n#### 选项 2: Cursor \u002F Windsurf\n编辑项目根目录或全局配置中的 `.cursor\u002Fmcp.json` 或 `~\u002F.config\u002Fwindsurf\u002Fmcp.json`：\n\n```json\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"kubectl-mcp-server\"]\n    }\n  }\n}\n```\n\n#### 选项 3: GitHub Copilot (VS Code)\n在 VS Code 的 `settings.json` 中添加：\n\n```json\n{\n  \"mcp\": {\n    \"servers\": {\n      \"kubernetes\": {\n        \"command\": \"npx\",\n        \"args\": [\"-y\", \"kubectl-mcp-server\"]\n      }\n    }\n  }\n}\n```\n\n> **注意**: 如果使用 `pip` 安装而非 `npx`，请将 `\"command\"` 改为 `\"python\"`，`\"args\"` 改为 `[\"-m\", \"kubectl_mcp_tool.mcp_server\"]`。\n\n### 第二步：开始对话\n\n重启 AI 助手后，直接在对话框中使用自然语言操作集群。\n\n**常用指令示例：**\n\n*   **故障排查**:\n    > \"为什么我的 nginx-deployment pod 一直在崩溃？\"\n    > \"诊断 default 命名空间下的网络连接问题。\"\n\n*   **资源管理**:\n    > \"列出 kube-system 命名空间下的所有 Pod。\"\n    > \"显示哪些 Pod 浪费了最多的资源。\"\n\n*   **应用部署**:\n    > \"创建一个包含 3 个副本的 Redis 部署。\"\n    > \"将 frontend 部署扩容到 5 个副本。\"\n\n*   **安全审计**:\n    > \"审计所有命名空间的 RBAC 权限。\"\n    > \"检查是否有使用特权模式运行的 Pod。\"\n\n*   **Helm 管理**:\n    > \"列出集群中所有的 Helm Release。\"\n    > \"从 Bitnami 仓库安装 Redis。\"\n\n### 第三步：CLI 独立测试 (可选)\n\n在不通过 AI 助手的情况下，你也可以直接在终端测试服务器功能：\n\n```bash\n# 查看服务器信息\nkubectl-mcp-server info\n\n# 列出所有可用工具\nkubectl-mcp-server tools\n\n# 直接调用工具获取 Pod 列表\nkubectl-mcp-server call get_pods '{\"namespace\": \"kube-system\"}'\n```","某电商平台的后端工程师在黑色星期五大促期间，突然收到告警：核心订单服务的 Pod 频繁崩溃重启，且伴随资源成本异常飙升，需要立即定位根因并优化。\n\n### 没有 kubectl-mcp-server 时\n- **命令记忆负担重**：工程师需在文档中反复查找复杂的 `kubectl` 参数（如 `top pods`、`describe`、`logs` 组合），排查效率极低。\n- **上下文切换频繁**：需要在终端、Grafana 仪表盘和 Helm 配置文件之间来回切换，难以将日志报错与资源配额直接关联。\n- **多步操作易出错**：手动执行“查看日志->分析镜像版本->回滚 Deployment\"的多步流程时，极易因输错命名空间或标签导致误操作。\n- **安全审计滞后**：无法实时通过自然语言询问“哪些 Pod 使用了特权模式”，只能事后编写脚本扫描，错失最佳修复窗口。\n- **协作沟通成本高**：向团队汇报问题时，需手动截图拼凑数据，无法即时生成可视化的资源拓扑图。\n\n### 使用 kubectl-mcp-server 后\n- **自然语言直达底层**：工程师直接对话 AI：“为什么订单服务 Pod 一直在重启？”，工具自动调用底层 API 聚合日志与事件，秒级返回根因是内存泄漏。\n- **一站式智能诊断**：只需一句“分析当前集群的成本浪费点并给出优化建议”，工具自动关联资源使用率与定价模型，直接输出缩容方案。\n- **原子化安全执行**：下达“将订单服务回滚到上一稳定版本”指令，工具自动解析意图、确认上下文并安全执行 Helm 回滚，消除人为手误风险。\n- **实时合规洞察**：随口提问“列出所有开启特权模式的容器”，工具即刻遍历集群并高亮风险项，实现动态安全审计。\n- **自动生成可视化报告**：请求“展示当前订单应用的资源拓扑”，工具直接渲染交互式仪表盘，让故障复盘会议一目了然。\n\nkubectl-mcp-server 将繁琐的 Kubernetes 命令行操作转化为直观的自然语言对话，让开发者像与资深 DevOps 专家交流一样轻松驾驭复杂集群。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Frohitg00_kubectl-mcp-server_8ec436b9.png","rohitg00","Rohit Ghumare","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Frohitg00_0db5b2cc.png","Principal Product Evangelist | DevRel | GDE @Google Cloud | CNCF Ambassador | Docker Captain | AWS CommunityBuilder | Ex - Solo.io, Cerbos, Oracle, Reliance Jio","Motia","London, UK",null,"ghumare64","https:\u002F\u002Fwww.devrelasservice.com\u002F","https:\u002F\u002Fgithub.com\u002Frohitg00",[86,90,94,98,102,106,110],{"name":87,"color":88,"percentage":89},"Python","#3572A5",81.1,{"name":91,"color":92,"percentage":93},"TypeScript","#3178c6",17,{"name":95,"color":96,"percentage":97},"Shell","#89e051",0.8,{"name":99,"color":100,"percentage":101},"JavaScript","#f1e05a",0.6,{"name":103,"color":104,"percentage":105},"HTML","#e34c26",0.2,{"name":107,"color":108,"percentage":109},"Dockerfile","#384d54",0.1,{"name":111,"color":112,"percentage":109},"Go Template","#00ADD8",859,168,"2026-04-04T16:01:26","MIT","Linux, macOS, Windows","未说明",{"notes":120,"python":121,"dependencies":122},"该工具是一个 Kubernetes MCP 服务器，需预先安装并配置 kubectl 以访问 Kubernetes 集群。支持通过 npx (Node.js) 或 pip (Python) 安装，也可使用 Docker 运行。若需使用交互式 UI 仪表盘功能，安装时需添加 [ui] 额外依赖。无 GPU 需求。","3.9+",[123,124,125],"kubectl","Node.js 14+ (可选，用于 npx)","Docker (可选)",[13,26,15,14],[128,129,130,131,132,133,134,135,136,137,138,139],"ai","deployment","devops","genai","kubernetes","kubernetes-cluster","llms","mcp","mcp-server","kubernetes-tools","npm","pypi","2026-03-27T02:49:30.150509","2026-04-06T07:13:24.777751",[143,148,153,157,162,166],{"id":144,"question_zh":145,"answer_zh":146,"source_url":147},16543,"在 Mac 上使用 Claude Desktop 时遇到 'Unexpected non-whitespace character after JSON' 错误怎么办？","这是一个已知问题，通常由 JSON 解析错误引起。维护者已提交修复补丁（commit 6beee4b）。请确保您使用的是最新版本的 kubectl-mcp-server。如果问题仍然存在，请检查日志中是否有额外的非 JSON 字符输出干扰了通信协议。","https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues\u002F9",{"id":149,"question_zh":150,"answer_zh":151,"source_url":152},16544,"配置 MCP 服务器时出现 'ModuleNotFoundError: No module named kubectl_mcp_tool' 错误如何解决？","这通常是因为配置文件中使用的 Python 解释器路径不正确或模块未在该环境中安装。请尝试修改 `mcp.json` 或 `config.json` 配置，使用通用的 `python` 命令并指定正确的参数。推荐配置如下：\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"python\",\n      \"args\": [\"-m\", \"kubectl_mcp_tool.minimal_wrapper\"],\n      \"env\": {\n        \"KUBECONFIG\": \"\u002Fpath\u002Fto\u002Fyour\u002F.kube\u002Fconfig\",\n        \"PATH\": \"\u002Fusr\u002Flocal\u002Fbin:\u002Fusr\u002Fbin:\u002Fbin:\u002Fusr\u002Fsbin:\u002Fsbin\",\n        \"MCP_LOG_FILE\": \"\u002Fpath\u002Fto\u002Flogs\u002Fdebug.log\",\n        \"MCP_DEBUG\": \"1\"\n      }\n    }\n  }\n}\n请确保将 `\u002Fpath\u002Fto\u002F...` 替换为您实际的配置文件路径和日志路径。","https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues\u002F14",{"id":154,"question_zh":155,"answer_zh":156,"source_url":152},16545,"遇到 'spawn python ENOENT' 错误意味着什么，如何修复？","该错误表示系统无法找到 `python` 可执行文件。这通常发生在配置文件中硬编码了不存在的 Python 路径，或者环境变量 `PATH` 中未包含 Python 安装目录。解决方法是在 MCP 配置文件的 `env` 字段中显式添加 `PATH` 环境变量，例如：`\"PATH\": \"\u002Fusr\u002Flocal\u002Fbin:\u002Fusr\u002Fbin:\u002Fbin:\u002Fusr\u002Fsbin:\u002Fsbin\"`，或者直接在使用 `command` 字段时指定 Python 的绝对路径（如 `\u002Fopt\u002Fanaconda3\u002Fbin\u002Fpython`），但需确保该路径下已安装 `kubectl-mcp-tool` 包。",{"id":158,"question_zh":159,"answer_zh":160,"source_url":161},16546,"为什么在命令行运行 `kubectl-mcp get pods` 会报错 'invalid choice: get'？","这是因为 `kubectl-mcp` 命令行工具的设计初衷是作为 MCP 服务器运行，而不是直接替代 `kubectl` 执行命令。目前该工具仅支持 `serve` 子命令来启动服务器（即 `kubectl-mcp serve`）。若要通过 AI 助手（如 Cursor）执行 Kubernetes 操作，请在 AI 客户端中配置好 MCP 服务器后，通过自然语言指令让 AI 调用工具，而不是直接在终端运行 `kubectl-mcp get ...`。","https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fissues\u002F5",{"id":163,"question_zh":164,"answer_zh":165,"source_url":152},16547,"如何在 Cursor 或 Claude 中正确配置 kubectl-mcp-server？","您需要在 AI 编辑器的 MCP 配置文件中添加服务器定义。对于 Cursor，编辑 `~\u002F.cursor\u002Fmcp.json`；对于 Claude Desktop，编辑 `~\u002F.config\u002Fclaude\u002Fmcp.json`（或相应配置文件）。配置内容示例如下：\n{\n  \"mcpServers\": {\n    \"kubernetes\": {\n      \"command\": \"python\",\n      \"args\": [\"-m\", \"kubectl_mcp_tool.minimal_wrapper\"],\n      \"env\": {\n        \"KUBECONFIG\": \"\u002FUsers\u002F您的用户名\u002F.kube\u002Fconfig\",\n        \"PATH\": \"\u002Fusr\u002Flocal\u002Fbin:\u002Fusr\u002Fbin:\u002Fbin:\u002Fusr\u002Fsbin:\u002Fsbin\"\n      }\n    }\n  }\n}\n配置完成后重启 AI 编辑器，并确保 `kubectl-mcp-tool` 已在当前 Python 环境中安装。",{"id":167,"question_zh":168,"answer_zh":169,"source_url":152},16548,"安装后如何验证 kubectl-mcp-tool 是否工作正常？","安装完成后，首先在终端运行 `kubectl-mcp --help` 确认命令可用且显示帮助信息。接着，运行 `kubectl-mcp serve` 启动服务器，观察是否有报错。如果在 AI 编辑器中配置后仍无法连接，可以手动测试包装器脚本：运行 `python -m kubectl_mcp_tool.minimal_wrapper` 查看是否有直接的 Python 错误输出。同时，检查 Kubernetes 配置文件（默认 `~\u002F.kube\u002Fconfig`）是否存在且有效，可以通过运行原生的 `kubectl get pods` 命令来预先验证 kubeconfig 的正确性。",[171,176,181,186,191,196,201,206,211,216,221,226,231,236,241,246,251,256,261,266],{"id":172,"version":173,"summary_zh":174,"released_at":175},98861,"v1.14.0","## v1.14.0 - Enhanced CLI + agent-browser v0.7 support\r\n\r\n### CLI Enhancements\r\n- **8 new subcommands**: `tools`, `resources`, `prompts`, `call`, `grep`, `info`, `context`, `doctor`\r\n- Structured error handling with actionable suggestions\r\n- Colorized output formatters with JSON mode support\r\n- Stdin support for call command arguments\r\n- NO_COLOR env var support\r\n\r\n### agent-browser v0.7 Support\r\n- 15 new environment variables for cloud providers, profiles, CDP, proxy settings\r\n- 7 new browser tools (26 total): `browser_connect_cdp`, `browser_install`, `browser_set_provider`, `browser_session_list`, `browser_session_switch`, `browser_open_with_headers`, `browser_set_viewport`\r\n- Retry with exponential backoff for transient errors\r\n- Enhanced `browser_open` with headers, session, and headed parameters\r\n\r\n### Tests\r\n- 18 new CLI tests\r\n- 12 new browser v0.7 tests (31 total browser tests)\r\n- All 187 tests passing\r\n\r\n### Usage Examples\r\n```\r\n# List all tools with descriptions\r\nkubectl-mcp-server tools -d\r\n\r\n# Search for pod-related tools\r\nkubectl-mcp-server grep \"*pod*\"\r\n\r\n# Call a tool directly\r\nkubectl-mcp-server call get_pods '{\"namespace\": \"kube-system\"}'\r\n\r\n# Check dependencies\r\nkubectl-mcp-server doctor\r\n```","2026-01-24T01:28:58",{"id":177,"version":178,"summary_zh":179,"released_at":180},98849,"v1.23.1","## 变更\n\n### 安全修复\n- 修复了 `networking.py` 中 `port-forward` 的 `os.system()` 命令注入问题，改用带列表参数的 `subprocess.Popen`\n- 修复了 `kubectl_apply` 中的临时文件泄漏问题，将 `os.unlink` 包装在 `try\u002Ffinally` 块中\n- 修复了 `kind_node_exec` 中的 `command.split()` 调用，改用 `shlex.split()`\n- 为 `exec_in_pod` 和 `port_forward` 添加了非破坏性保护机制\n- 限制了 `kubectl_generic` 的白名单：`config` 仅允许安全的子命令，`auth` 仅允许 `can-i`\n- 修复了 HTTP 处理器中的硬编码版本字符串\n\n### DRY 整合\n- 将 `_get_kubectl_context_args` 从 10 处重复实现整合为 `k8s_config.py` 中的单一来源\n- 提取了共享的 Helm 仓库添加\u002F更新辅助函数\n- 创建了 `_cli_utils.py`，包含缓存的 CLI 可用性检查和通用的 subprocess 运行器\n- 在所有工具文件中统一了非破坏性模式的编写规范\n\n### 移除死代码\n- 移除了 7 个文件中的未使用导入\n- 移除了 `safety.py` 中未使用的 `check_safety_mode` 装饰器和 `is_operation_allowed` 函数\n- 移除了 `mcp_server.py` 中的自动 pip 安装反模式\n\n### 代码质量\n- 修复了 `networking.py` 和 `cost.py` 中的裸 `except:` 子句\n- 修复了 Pod 推荐列表中的 `None` 值\n- 在 FastMCP 导入中使用 `from err` 正确地进行异常链式传递\n- 更新了测试套件：469 个测试通过\n\n## 安装\n\n```bash\npip install kubectl-mcp-server==1.23.1\n# 或者\nnpx kubectl-mcp-server@1.23.1\n```","2026-02-04T00:33:05",{"id":182,"version":183,"summary_zh":184,"released_at":185},98850,"v1.23.0","## 修复：集群内 Kubernetes 配置\n\n解决了当服务器部署在 Kubernetes 集群内部时出现的 `HTTPConnectionPool(host='localhost', port=80): Max retries exceeded` 错误（#61）。\n\n### 变更内容\n\n- **修复配置加载** — `_patched_load_kube_config` 不再对带有显式参数（config_file、context、client_configuration）的调用进行短路处理，这曾导致 Kubernetes 客户端无法加载正确的 API 服务器地址。\n- **集群内回退机制** — 当容器中不存在 kubeconfig 文件时，kubeconfig 提供者现在会自动回退到集群内配置。\n- **可操作的错误提示** — 在既没有 kubeconfig 也没有集群内配置可用时，抛出带有明确解决步骤的 `ProviderError`，而不是静默地回退到 `localhost:80`。\n- **部署清单** — 添加了 `MCP_K8S_PROVIDER=in-cluster` 和 `KUBERNETES_SERVICE_PORT=443` 环境变量。\n\n### 部署\n\n对于集群内部署，请确保设置以下环境变量：\n\n```yaml\nenv:\n  - name: MCP_K8S_PROVIDER\n    value: \"in-cluster\"\n  - name: KUBERNETES_SERVICE_HOST\n    value: \"kubernetes.default.svc\"\n  - name: KUBERNETES_SERVICE_PORT\n    value: \"443\"\n```\n\n参考清单：[`deploy\u002Fkubernetes\u002F`](https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Ftree\u002Fmain\u002Fdeploy\u002Fkubernetes)\n\n### 安装\n\n```bash\npip install kubectl-mcp-server==1.23.0\n# 或\nnpx kubectl-mcp-server@1.23.0\n# 或\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:1.23.0\n```","2026-02-03T10:48:58",{"id":187,"version":188,"summary_zh":189,"released_at":190},98862,"v1.13.0","## What's New\n\n### MCP-UI Interactive Dashboards (6 new tools)\n\nThis release adds support for [MCP-UI](https:\u002F\u002Fgithub.com\u002FMCP-UI-Org\u002Fmcp-ui) interactive HTML dashboards in compatible hosts (Goose, LibreChat, Nanobot).\n\n| Tool | Description |\n|------|-------------|\n| `show_pod_logs_ui` | Interactive log viewer with search and filtering |\n| `show_pods_dashboard_ui` | Pods table with status indicators |\n| `show_resource_yaml_ui` | YAML viewer with syntax highlighting |\n| `show_cluster_overview_ui` | Cluster dashboard (nodes, namespaces, workloads) |\n| `show_events_timeline_ui` | Events timeline with severity filtering |\n| `render_k8s_dashboard_screenshot` | Render any dashboard as PNG screenshot |\n\n### Features\n\n- **Dark theme**: Catppuccin-style UI optimized for terminals\n- **Optional dependency**: Install with `pip install kubectl-mcp-server[ui]`\n- **Graceful fallback**: Returns JSON data when MCP-UI not supported\n- **Screenshot rendering**: Works with agent-browser for universal compatibility\n\n### Compatibility\n\n| Host | MCP-UI Support | Fallback |\n|------|----------------|----------|\n| Goose | ✅ Full | - |\n| LibreChat | ✅ Full | - |\n| Nanobot | ✅ Full | - |\n| Claude Desktop | ❌ | JSON + Screenshot |\n| Cursor | ❌ | JSON + Screenshot |\n\n## Installation\n\n```bash\n# Standard install (127 tools)\npip install kubectl-mcp-server\n\n# With MCP-UI support\npip install kubectl-mcp-server[ui]\n\n# Via npm\nnpx kubectl-mcp-server@1.13.0\n```\n\n## Full Changelog\nhttps:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.12.0...v1.13.0","2026-01-23T12:28:18",{"id":192,"version":193,"summary_zh":194,"released_at":195},98851,"v1.22.0","## 新增内容\n\n### kubectl-mcp-app - 8 个交互式 UI 仪表板\n\n新增了一个独立的 npm 包 `kubectl-mcp-app`，它基于 MCP ext-apps SDK，为 Kubernetes 管理提供了美观且交互性强的 UI 仪表板。\n\n**安装：**\n```bash\n# 通过 npm 安装\nnpm install -g kubectl-mcp-app\n\n# 或者使用 npx（无需安装）\nnpx kubectl-mcp-app\n```\n\n**Claude Desktop 配置：**\n```json\n{\n  \"mcpServers\": {\n    \"kubectl-app\": {\n      \"command\": \"npx\",\n      \"args\": [\"kubectl-mcp-app\"]\n    }\n  }\n}\n```\n\n**8 个交互式 UI 工具：**\n\n| 工具 | 描述 |\n| ---- | ----------- |\n| `k8s-pods` | 带有筛选、排序和状态指示器的交互式 Pod 查看器 |\n| `k8s-logs` | 支持语法高亮和搜索的实时日志查看器 |\n| `k8s-deploy` | 显示部署滚动更新状态、扩缩容和回滚操作的仪表板 |\n| `k8s-helm` | 提供升级\u002F回滚功能的 Helm 发布管理器 |\n| `k8s-cluster` | 展示节点健康状况和资源指标的集群概览 |\n| `k8s-cost` | 具备浪费检测与优化建议的成本分析工具 |\n| `k8s-events` | 支持按类型筛选和分组的事件时间线 |\n| `k8s-network` | 展示 Service、Pod 和 Ingress 的网络拓扑图 |\n\n**特性：**\n- 🎨 支持深色\u002F浅色主题\n- 📊 实时数据可视化  \n- 🖱️ 支持交互式操作（扩缩容、重启、删除）\n- 🔗 与 kubectl-mcp-server 无缝集成\n- 📦 使用 TypeScript + React 19 + Vite 构建\n- 🎯 单文件 HTML 打包（每个约 220KB）\n- ✅ 27 个测试，覆盖率达到 100%\n\n## 完整变更日志\nhttps:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.21.0...v1.22.0","2026-02-02T10:45:40",{"id":197,"version":198,"summary_zh":199,"released_at":200},98863,"v1.12.0","## What's New in v1.12.0\n\n### SSE Transport Fix\n- Fixed `'FastMCP' object has no attribute 'run_sse_async'` error when using FastMCP 3.0.0b1\n- Now uses `create_sse_app()` from `fastmcp.server.http` module\n- SSE endpoints: GET `\u002Fsse` (events), POST `\u002Fmessages\u002F` (messages)\n\n### In-Cluster Kubernetes Config Support\n- Added `k8s_config.py` module that automatically handles Kubernetes authentication\n- Tries in-cluster config first when running inside a Kubernetes pod\n- Falls back to kubeconfig file for local development\n- Enables kubectl-mcp-server to work seamlessly with Kubernetes service accounts\n\n### kagent Integration\n- Added `deploy\u002Fkagent\u002Fremotemcpserver.yaml` for kagent RemoteMCPServer CRD\n- Added `deploy\u002Fkagent\u002Fagent-kubectl-mcp.yaml` for kagent Agent CRD  \n- Successfully tested with kagent v0.7.9 - all 121 tools discovered\n\n## Installation\n\n### PyPI\n\\`\\`\\`bash\npip install kubectl-mcp-server==1.12.0\n\\`\\`\\`\n\n### npm\n\\`\\`\\`bash\nnpx kubectl-mcp-server@1.12.0\n\\`\\`\\`\n\n### Docker\n\\`\\`\\`bash\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:1.12.0\n\\`\\`\\`\n\n## Full Changelog\nhttps:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.11.0...v1.12.0","2026-01-22T22:47:37",{"id":202,"version":203,"summary_zh":204,"released_at":205},98852,"v1.21.0","## 新增内容\n\n### vCluster（vind）支持 - 14 个工具\n使用 vCluster CLI 管理虚拟 Kubernetes 集群：\n- `vind_detect_tool`、`vind_list_clusters_tool`、`vind_status_tool`\n- `vind_create_cluster_tool`、`vind_delete_cluster_tool`\n- `vind_pause_tool`、`vind_resume_tool`、`vind_connect_tool`、`vind_disconnect_tool`\n- `vind_upgrade_tool`、`vind_describe_tool`、`vind_get_kubeconfig_tool`\n- `vind_logs_tool`、`vind_platform_start_tool`\n\n### kind（Kubernetes IN Docker）支持 - 32 个工具\n全面的本地 Kubernetes 集群管理：\n\n**核心操作（15 个工具）：**\n- 集群生命周期：创建、删除、列出、查看信息\n- 镜像管理：加载、构建节点镜像\n- Kubeconfig：获取、设置、导出\n\n**v1.21.0 中新增（17 个工具）：**\n- **配置**：`kind_config_validate`、`kind_config_generate`、`kind_config_show`、`kind_available_images`\n- **Registry**：`kind_registry_create`、`kind_registry_connect`、`kind_registry_status`\n- **节点管理**：`kind_node_exec`、`kind_node_logs`、`kind_node_inspect`、`kind_node_restart`\n- **网络**：`kind_network_inspect`、`kind_port_mappings`、`kind_ingress_setup`\n- **诊断**：`kind_cluster_status`、`kind_images_list`、`kind_provider_info`\n\n### 增强的 Agent 技能 - 26 项技能\n所有 26 项 Kubernetes 技能均已更新，以符合 [agenstskills.com](https:\u002F\u002Fagenstskills.com) 规范：\n- 增强 YAML 前置元数据（许可证、作者、版本、工具、类别）\n- “适用场景”激活触发章节\n- 优先级规则表（CRITICAL\u002FHIGH\u002FMEDIUM\u002FLOW）\n- 常见操作快速参考表\n\n**新增参考文档：**\n- `k8s-troubleshoot\u002Freferences\u002FDECISION-TREE.md`、`COMMON-ERRORS.md`\n- `k8s-deploy\u002Freferences\u002FSTRATEGIES.md`\n- `k8s-helm\u002Freferences\u002FCHART-STRUCTURE.md`\n- `k8s-kind\u002Freferences\u002FCONFIGS.md`\n- `k8s-vind\u002Freferences\u002FWORKFLOWS.md`\n\n**新增示例清单：**\n- `k8s-deploy\u002Fexamples\u002Fcanary-rollout.yaml`、`blue-green.yaml`、`hpa-deployment.yaml`\n- `k8s-autoscaling\u002Fexamples\u002Fhpa-cpu.yaml`、`keda-scaledobject.yaml`\n\n## 工具数量\n- **总工具数**：270+（原为 235+）\n- **vind 工具**：14 个（新增）\n- **kind 工具**：32 个（从 15 个扩展而来）\n- **Agent 技能**：26 项\n\n## 安装\n\n```bash\n# pip\npip install kubectl-mcp-server==1.21.0\n\n# npm\nnpx kubectl-mcp-server@1.21.0\n\n# Docker\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:1.21.0\n```\n\n## 完整变更日志\nhttps:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.19.3...v1.21.0","2026-01-30T12:27:27",{"id":207,"version":208,"summary_zh":209,"released_at":210},98853,"v1.19.3","## 变更内容\n- 增加了对 GitHub Container Registry (ghcr.io) 的发布支持\n- Docker 镜像现在同时在 Docker Hub 和 GitHub Packages 上提供\n\n## Docker 镜像\n```bash\n# Docker Hub\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:1.19.3\n\n# GitHub Container Registry\ndocker pull ghcr.io\u002Frohitg00\u002Fkubectl-mcp-server:1.19.3\n```\n\n## 从 GitHub Packages 安装\n```bash\n# npm\nnpm install @rohitg00\u002Fkubectl-mcp-server --registry=https:\u002F\u002Fnpm.pkg.github.com\n\n# Python（使用发布中的 wheel 包）\npip install https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases\u002Fdownload\u002Fv1.19.3\u002Fkubectl_mcp_server-1.19.3-py3-none-any.whl\n```","2026-01-27T11:49:22",{"id":212,"version":213,"summary_zh":214,"released_at":215},98854,"v1.19.2","## 变更内容\n- 为 npm 包添加了 GitHub Packages 发布功能\n- 该包现已在 GitHub Packages 上以 `@rohitg00\u002Fkubectl-mcp-server` 的名称提供\n\n## 从 GitHub Packages 安装\n```bash\nnpm install @rohitg00\u002Fkubectl-mcp-server --registry=https:\u002F\u002Fnpm.pkg.github.com\n```","2026-01-27T11:45:34",{"id":217,"version":218,"summary_zh":219,"released_at":220},98855,"v1.19.1","## 变更内容\n- 将 wheel 文件上传至 GitHub 发布页面，以便直接使用 pip 安装\n- 更新了文档，添加了通过 GitHub 发布页面安装的说明\n\n## 从 GitHub 发布页面安装\n```bash\npip install https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Freleases\u002Fdownload\u002Fv1.19.1\u002Fkubectl_mcp_server-1.19.1-py3-none-any.whl\n```","2026-01-27T11:42:31",{"id":222,"version":223,"summary_zh":224,"released_at":225},98864,"v1.11.0","## What's New\n\n### 🌐 Browser Automation (19 tools)\nOptional module using [agent-browser](https:\u002F\u002Fgithub.com\u002Fvercel-labs\u002Fagent-browser) CLI for browser automation within Kubernetes workflows.\n\n**Enable with:** `MCP_BROWSER_ENABLED=true`\n\n**Core Tools:**\n- `browser_open`, `browser_snapshot`, `browser_click`, `browser_fill`\n- `browser_screenshot`, `browser_get_text`, `browser_wait`, `browser_close`\n\n**Kubernetes-Specific:**\n- `browser_test_ingress` - Test ingress URLs and capture screenshots\n- `browser_screenshot_grafana` - Capture Grafana dashboards\n- `browser_screenshot_argocd` - Capture ArgoCD application status\n- `browser_open_cloud_console` - Generate URLs for AWS\u002FGCP\u002FAzure consoles\n\n**Utilities:**\n- `browser_health_check`, `browser_form_submit`, `browser_pdf_export`\n- `browser_session_save`, `browser_session_load`\n\n### 📦 Package Renaming\n- **Primary package:** `kubectl-mcp-server` (new name)\n- **Legacy alias:** `kubectl-mcp-tool` (for backward compatibility)\n- Both packages are published to PyPI\n\n**Install:**\n```bash\n# Recommended (new name)\npip install kubectl-mcp-server\n\n# Legacy (still works)\npip install kubectl-mcp-tool\n```\n\n### 📊 Stats\n- **140 tools** (121 core + 19 browser tools when enabled)\n- **8 resources**\n- **8 prompts**\n- **187 tests** passing\n\n### Full Changelog\nhttps:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.10.0...v1.11.0","2026-01-22T12:26:02",{"id":227,"version":228,"summary_zh":229,"released_at":230},98856,"v1.19.0","## v1.19.0 新增内容\n\n### 新特性\n\n#### 多集群操作（新增4个工具）\n- **`multi_cluster_query`**：同时查询多个集群中的资源\n- **`multi_cluster_health`**：一次性检查所有集群的健康状态\n- **`multi_cluster_pod_count`**：获取所有集群的Pod数量，并按状态细分\n\n#### Kubeconfig 自动检测\n- **`enable_kubeconfig_watching`**：自动检测kubeconfig文件的变化（适用于云CLI凭证更新场景）\n- **`disable_kubeconfig_watching`**：禁用监听功能\n- **`get_server_config_status`**：查看当前服务器配置，包括监听状态\n\n#### 无状态模式\n- 环境变量 **`MCP_STATELESS_MODE`**：无需缓存API客户端即可运行（非常适合无服务器环境）\n- **`set_stateless_mode`** \u002F **`is_stateless_mode`**：通过编程方式切换无状态模式\n\n#### Pod 管理\n- **`run_pod`**：直接运行容器镜像（等同于`kubectl run`）\n\n### 代码质量改进\n- 创建了共享的`utils.py`模块，移除了约474行重复代码\n- 从工具文件中删除了31条冗余注释\n- 为`check_interval`参数添加了输入校验\n- 简化了模块文档字符串\n- 所有47个测试均通过\n\n### 文件变更\n- 修改了20个文件\n- 新增：`kubectl_mcp_tool\u002Ftools\u002Futils.py`\n\n## 安装\n\n```bash\n# PyPI\npip install kubectl-mcp-server==1.19.0\n\n# npm\nnpm install kubectl-mcp-server@1.19.0\n```\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.18.0...v1.19.0","2026-01-25T18:32:41",{"id":232,"version":233,"summary_zh":234,"released_at":235},98857,"v1.18.0","## 新增内容\n\n### 代理技能库（24项技能）\n\n根据 [Agent Skills](https:\u002F\u002Fagentskills.io) 规范，新增了全面的 Kubernetes 技能，适用于 AI 编码代理。\n\n**分类：**\n| 分类 | 技能 |\n|----------|--------|\n| **核心资源** | k8s-core, k8s-networking, k8s-storage |\n| **工作负载** | k8s-deploy, k8s-operations, k8s-helm |\n| **可观测性** | k8s-diagnostics, k8s-troubleshoot, k8s-incident |\n| **安全** | k8s-security, k8s-policy, k8s-certs |\n| **GitOps** | k8s-gitops, k8s-rollouts |\n| **扩展性** | k8s-autoscaling, k8s-cost, k8s-backup |\n| **多集群** | k8s-multicluster, k8s-capi, k8s-kubevirt |\n| **网络** | k8s-service-mesh, k8s-cilium |\n| **工具** | k8s-browser, k8s-cli |\n\n**安装：**\n```bash\n# 将所有技能复制到 Claude\ncp -r kubernetes-skills\u002Fclaude\u002F* ~\u002F.claude\u002Fskills\u002F\n\n# 或使用 SkillKit 转换为其他代理\nnpm install -g skillkit\nskillkit translate kubernetes-skills\u002Fclaude --to cursor --output .cursor\u002Frules\u002F\n```\n\n### 增强的提供者模块\n\n新增 `providers.py` 模块，用于更好地管理多集群：\n\n- **单例模式**：`KubernetesProvider.get_instance()` 提供共享的提供者实例\n- **API 客户端缓存**：为每个上下文缓存客户端，以提升性能\n- **上下文验证**：抛出 `UnknownContextError` 异常，并列出可用的上下文\n- **环境变量**：\n  - `MCP_K8S_PROVIDER`：指定 kubeconfig、集群内或单个提供者\n  - `MCP_K8S_CONTEXT`：单个提供者的默认上下文\n  - `MCP_K8S_QPS`：API 速率限制（默认：100）\n  - `MCP_K8S_BURST`：API 突发限制（默认：200）\n  - `MCP_K8S_TIMEOUT`：请求超时时间（默认：30 秒）\n\n### 完整变更日志\n\n- 新增 24 项代理技能，覆盖超过 224 种工具\n- 新增 `providers.py` 模块，增强提供者管理功能\n- 将提供者集成到 `k8s_config.py` 中，并支持回退机制\n- 更新文档，添加技能安装指南\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.17.0...v1.18.0","2026-01-25T14:56:42",{"id":237,"version":238,"summary_zh":239,"released_at":240},98865,"v1.10.0","## Security-first release with MCP Authorization (RFC 9728) and kagent integration.\n\n### MCP Authorization (RFC 9728)\n- **OAuth 2.0 Protected Resource Metadata** for enterprise security\n- **JWT verification** via JWKS endpoints\n- **Enterprise IdP support**: Okta, Auth0, Keycloak, Microsoft Entra ID, Google OAuth\n- **Fine-grained scopes** for access control:\n  | Scope | Description |\n  |-------|-------------|\n  | `mcp:read` | Read-only operations |\n  | `mcp:write` | Write operations |\n  | `mcp:admin` | Administrative operations |\n  | `mcp:tools` | General tool access |\n  | `mcp:helm` | Helm operations |\n  | `mcp:diagnostics` | Diagnostic operations |\n  | `mcp:networking` | Network operations |\n  | `mcp:storage` | Storage operations |\n  | `mcp:security` | Security operations |\n  | `mcp:cost` | Cost analysis operations |\n\n### Environment Variables\n```bash\nMCP_AUTH_ENABLED=true\nMCP_AUTH_ISSUER_URL=https:\u002F\u002Fyour-idp.com\nMCP_AUTH_JWKS_URI=https:\u002F\u002Fyour-idp.com\u002F.well-known\u002Fjwks.json\nMCP_AUTH_AUDIENCE=kubectl-mcp-server\nMCP_AUTH_REQUIRED_SCOPES=mcp:read,mcp:tools\n```\n\n### kagent Integration\n- **ToolServer manifest** for Kubernetes-native MCP server deployment\n- **Agent manifest** for AI agent orchestration with kagent\n- Deploy to Kubernetes with full MCP protocol support\n\n### Testing\n- **167 tests passing** (up from 138)\n- Added 29 new auth module tests\n- Comprehensive scope and JWT verification tests\n\n### Installation\n```bash\n# npm\u002Fnpx\nnpx kubectl-mcp-server\n\n# pip\npip install kubectl-mcp-tool==1.10.0\n\n# Docker\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:v1.10.0\n```","2026-01-22T10:48:44",{"id":242,"version":243,"summary_zh":244,"released_at":245},98858,"v1.17.0","## v1.17.0 - MCP Server Enterprise Integration\n\nThis release integrates enterprise-readiness modules directly into the MCP server, building on the PRD implementation from previous releases.\n\n### MCP Server Integration\n\n**HTTP Endpoints Added:**\n- `\u002Fstats` - Real-time server statistics and tool call metrics\n- `\u002Fmetrics` - Prometheus-compatible metrics (when available)\n- `\u002Fsafety` - Safety mode status and configuration\n\n**CLI Parameters:**\n- `--config \u003Cfile>` - Load configuration from TOML file\n- `--read-only` - Enable read-only safety mode\n- `--disable-destructive` - Disable destructive operations only\n\n**Configuration Features:**\n- TOML configuration file support\n- SIGHUP signal handler for runtime config reload\n- CLI flags take precedence over config file settings\n- Custom prompts loading from config\n\n### Safety Modes\n- **Normal**: All operations allowed\n- **Read-Only**: Only read operations (get, list, describe)\n- **Disable-Destructive**: Block destructive operations (delete, drain, cordon)\n\n### Observability\n- StatsCollector for tool call metrics\n- Prometheus metrics integration (optional)\n- OpenTelemetry tracing support (optional)\n\n### Example Usage\n\n```bash\n# Start with config file\nkubectl-mcp-server serve --config \u002Fetc\u002Fkubectl-mcp\u002Fconfig.toml\n\n# Start in read-only mode\nkubectl-mcp-server serve --read-only\n\n# Disable destructive operations only\nkubectl-mcp-server serve --disable-destructive\n```\n\n### Installation\n- **pip**: `pip install kubectl-mcp-server==1.17.0`\n- **npm**: `npx kubectl-mcp-server@1.17.0`\n- **Docker**: `docker pull rohitghumare64\u002Fkubectl-mcp-server:1.17.0`\n\n### Full Changelog\nhttps:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.16.0...v1.17.0","2026-01-25T01:50:55",{"id":247,"version":248,"summary_zh":249,"released_at":250},98859,"v1.16.0","## What's New\n\nAdded **60 new tools** across 6 advanced Kubernetes ecosystem toolsets, bringing the total from 164 to 224 tools.\n\n### New Ecosystem Toolsets\n\n| Toolset | Tools | Description |\n|---------|-------|-------------|\n| **KEDA** | 7 | Event-driven autoscaling (ScaledObjects, ScaledJobs, TriggerAuthentications) |\n| **Cilium\u002FHubble** | 8 | eBPF-powered networking & observability (NetworkPolicies, Hubble flows) |\n| **Argo Rollouts\u002FFlagger** | 11 | Progressive delivery (Canary, Blue-Green, AnalysisRuns) |\n| **Cluster API** | 11 | Declarative cluster lifecycle management |\n| **KubeVirt** | 13 | VM management on Kubernetes (start, stop, migrate, pause) |\n| **Istio\u002FKiali** | 10 | Service mesh observability (VirtualServices, DestinationRules, Gateways) |\n\n### Improvements\n- Improved error handling in CAPI kubeconfig lookup (properly propagates non-NotFound errors)\n- Fixed Argo Rollouts full promotion to use `promoteFull=True`\n\n### Installation\n\n```bash\n# pip\npip install kubectl-mcp-server==1.16.0\n\n# npm\nnpx kubectl-mcp-server@1.16.0\n\n# Docker\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:1.16.0\n```\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.15.0...v1.16.0","2026-01-24T21:37:49",{"id":252,"version":253,"summary_zh":254,"released_at":255},98866,"v1.9.0","## Major refactoring to improve maintainability and code organization.\n\n### Architecture Changes\n- **Modularized mcp_server.py** from ~7,400 lines to ~450 lines\n- **Extracted 121 tools** into 11 category modules in `tools\u002F`:\n  | Module | Description | Tools |\n  |--------|-------------|-------|\n  | `pods.py` | Pod management and diagnostics | 11 |\n  | `deployments.py` | Deployments, StatefulSets, DaemonSets, Jobs | 10 |\n  | `core.py` | Namespaces, ConfigMaps, Secrets, Events | 6 |\n  | `cluster.py` | Context\u002Fcluster management | 11 |\n  | `networking.py` | Services, Ingress, NetworkPolicies | 8 |\n  | `storage.py` | PVCs, StorageClasses, PVs | 3 |\n  | `security.py` | RBAC, ServiceAccounts, PodSecurity | 10 |\n  | `helm.py` | Complete Helm v3 operations | 37 |\n  | `operations.py` | kubectl apply\u002Fpatch\u002Fdescribe\u002Frollout | 15 |\n  | `diagnostics.py` | Metrics and namespace comparison | 3 |\n  | `cost.py` | Resource optimization and cost analysis | 9 |\n\n- **Extracted resources** into `resources\u002Fresources.py` (8 MCP resources)\n- **Extracted prompts** into `prompts\u002Fprompts.py` (8 MCP prompts)\n- **Added** `utils\u002Fhelpers.py` for shared utilities\n\n### Testing\n- **138 tests passing**\n- Added comprehensive tool registration tests\n- Verifies all 121 tools are registered correctly\n\n### Documentation\n- Updated README with modular architecture diagram\n- Updated tool count to 121\n\n### Installation\n```bash\n# npm\u002Fnpx\nnpx kubectl-mcp-server\n\n# pip\npip install kubectl-mcp-tool==1.9.0\n\n# Docker\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:v1.9.0\n```","2026-01-22T01:06:07",{"id":257,"version":258,"summary_zh":259,"released_at":260},98867,"v1.8.0","## What's New in v1.8.0\n\n### MCP Resources (14 new resources)\nAccess Kubernetes data as browsable resources with FastMCP 3:\n\n| Resource URI | Description |\n|--------------|-------------|\n| `kubeconfig:\u002F\u002Fcontexts` | List all kubectl contexts |\n| `kubeconfig:\u002F\u002Fcurrent-context` | Current active context |\n| `namespace:\u002F\u002Fcurrent` | Current namespace |\n| `namespace:\u002F\u002Flist` | List all namespaces |\n| `cluster:\u002F\u002Finfo` | Cluster info with version & nodes |\n| `cluster:\u002F\u002Fnodes` | Detailed node information |\n| `cluster:\u002F\u002Fversion` | Kubernetes version |\n| `cluster:\u002F\u002Fapi-resources` | Available API resources |\n| `manifest:\u002F\u002Fdeployments\u002F{ns}\u002F{name}` | Deployment YAML |\n| `manifest:\u002F\u002Fservices\u002F{ns}\u002F{name}` | Service YAML |\n| `manifest:\u002F\u002Fpods\u002F{ns}\u002F{name}` | Pod YAML |\n| `manifest:\u002F\u002Fconfigmaps\u002F{ns}\u002F{name}` | ConfigMap YAML |\n| `manifest:\u002F\u002Fsecrets\u002F{ns}\u002F{name}` | Secret YAML (data masked) |\n| `manifest:\u002F\u002Fingresses\u002F{ns}\u002F{name}` | Ingress YAML |\n\n### MCP Prompts (8 new prompts)\nPre-built workflow prompts for common Kubernetes operations:\n\n| Prompt | Description |\n|--------|-------------|\n| `troubleshoot_workload` | Comprehensive pod\u002Fdeployment troubleshooting guide |\n| `deploy_application` | Step-by-step deployment workflow |\n| `security_audit` | Security scanning and RBAC analysis |\n| `cost_optimization` | Resource optimization and cost analysis |\n| `disaster_recovery` | Backup and recovery planning |\n| `debug_networking` | Network debugging for services |\n| `scale_application` | Scaling guide with HPA\u002FVPA best practices |\n| `upgrade_cluster` | Kubernetes cluster upgrade planning |\n\n### Comprehensive Test Suite\n- **135 tests** covering all tools, resources, and prompts\n- pytest configuration with fixtures\n- Development dependencies in `requirements-dev.txt`\n\n### Documentation\n- Added Development & Testing section to README\n- Test running instructions\n- Code quality commands\n\n### Fixes\n- FastMCP 3 compatibility improvements\n- Fixed auto-install version for beta package\n\n## Installation\n\n```bash\n# npm\nnpm install -g kubectl-mcp-server\n\n# pip\npip install kubectl-mcp-tool\n\n# Docker\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:v1.8.0\n```\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.7.0...v1.8.0","2026-01-21T21:25:43",{"id":262,"version":263,"summary_zh":264,"released_at":265},98860,"v1.15.0","## Multi-Cluster Support with Context Parameter\n\nThis release adds comprehensive multi-cluster support by introducing an optional `context` parameter to all cluster-interacting tools. Target any Kubernetes cluster without switching contexts!\n\n### New Features\n\n- **Context Parameter for All Tools**: All 131 tools now accept an optional `context` parameter to target specific clusters\n- **Cross-Cluster Operations**: Run commands against any cluster in your kubeconfig without switching contexts\n- **Response Enrichment**: Every response now includes `\"context\": \"production\"` or `\"context\": \"current\"` for clarity\n\n### Usage Examples\n\n```bash\n# Target a specific cluster context\nkubectl-mcp-server call get_pods '{\"namespace\": \"default\", \"context\": \"production\"}'\n\n# Get deployments from staging\nkubectl-mcp-server call get_deployments '{\"namespace\": \"app\", \"context\": \"staging\"}'\n\n# Install Helm chart to production cluster\nkubectl-mcp-server call install_helm_chart '{\"name\": \"redis\", \"chart\": \"bitnami\u002Fredis\", \"namespace\": \"cache\", \"context\": \"production\"}'\n```\n\n**Natural language with AI assistants:**\n- \"List pods in the production cluster\"\n- \"Get deployments from staging context\"\n- \"Show logs from the api-pod in the dev cluster\"\n\n### Technical Details\n\n- Enhanced `k8s_config.py` with context-aware client creation (`get_k8s_client(context)`, `get_apps_client(context)`)\n- Helper functions: `_get_kubectl_context_args(context)`, `_get_helm_context_args(context)`\n- If `context` is omitted, tools use the current kubectl context (backward compatible)\n\n### Stats\n\n- **131 Core Tools** (up from 127)\n- **216 Tests Passing**\n- **26 Browser Tools** (optional)\n- **6 UI Tools** (optional)\n\n### Installation\n\n```bash\n# npm (recommended)\nnpx kubectl-mcp-server@1.15.0\n\n# pip\npip install kubectl-mcp-server==1.15.0\n\n# Docker\ndocker pull rohitghumare64\u002Fkubectl-mcp-server:1.15.0\n```\n\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.14.0...v1.15.0","2026-01-24T14:36:14",{"id":267,"version":268,"summary_zh":269,"released_at":270},98848,"v1.24.0","## 3D 集群拓扑查看器\n\n\u003Cimg width=\"823\" height=\"366\" alt=\"截图 2026-02-20 22:26:21\" src=\"https:\u002F\u002Fgithub.com\u002Fuser-attachments\u002Fassets\u002F5d26057d-feb1-4a72-ace2-d2f4ae2ef214\" \u002F>\n\n基于 Three.js 构建的全新交互式 3D Kubernetes 集群拓扑 UI（`k8s-3d-topology`）。将您的整个集群可视化为可导航的 3D 场景。\n\n### 功能特性\n- **15 种不同的 3D 网格类型**，用于表示 K8s 资源（Pod、Deployment、ReplicaSet、Service、Ingress、Node、StatefulSet、DaemonSet、ConfigMap、Secret、PVC、HPA、NetworkPolicy、Job、CronJob）\n- **力导向布局**，并按命名空间聚类\n- **关系边线**，展示所有权、网络、存储和配置连接，并带有动画流效果\n- **拖拽重定位**节点，更新后位置保持不变\n- **点击查看详情**，集成 `kubectl describe`\n- **SVG 小地图**，用于概览图结构\n- **命名空间、资源类型及搜索过滤**\n- **深色主题**，搭配 Kubernetes 蓝色点缀、阴影、雾效和发光效果\n- **LRU 纹理缓存**（128 个条目），以高效利用 GPU 内存\n\n### 新增文件（共 16 个文件，2,494 行）\n- `App.tsx` - 主拓扑应用，包含主题\u002F过滤\u002F选中状态\n- `ClusterScene.tsx` - Three.js 场景，配备轨道控件、光线投射和阴影\n- `FilterBar.tsx` - 命名空间、资源类型及搜索过滤组件\n- `InspectorSidebar.tsx` - 资源详情页，集成 kubectl describe\n- `Minimap.tsx` - 图结构的 SVG 概览\n- `useClusterData.ts` - 使用 MCP 工具调用 Pod、Deployment、Service、Ingress、ReplicaSet 和 Node 数据\n- `meshFactory.ts` - 15 种不同的 3D 网格生成器，配备 LRU 标签缓存\n- `layoutEngine.ts` - 力导向布局，支持命名空间聚类\n- `k8sRelationships.ts` - 检测所有权、网络、存储和配置关系边线\n- `types.ts` - TypeScript 接口，定义图节点、边和资源类型\n- `constants.ts` - 共享的资源类型颜色表\n\n### 同时包含\n- 适用于 GitHub Pages 的着陆页，采用 Kubernetes 蓝色单色主题\n- CNCF Landscape 列表及经验证的媒体引用\n\n### 安装方法\n```bash\nnpm install -g kubectl-mcp-app\n# 或\nnpx kubectl-mcp-app\n```\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Frohitg00\u002Fkubectl-mcp-server\u002Fcompare\u002Fv1.23.1...v1.24.0","2026-02-20T16:53:08"]