InfraClaude
An MCP server that enables Claude to manage infrastructure across Kubernetes, Docker, Prometheus, and Terraform through natural language. It provides over 42 specialized tools with a safety-first design, including risk-based command classification and audit logging.
README
InfraClaude
Give Claude superpowers over your infrastructure — MCP Server for Kubernetes, Docker, Terraform, Prometheus, and Security tools.
InfraClaude is an MCP (Model Context Protocol) server that exposes infrastructure management tools to Claude. Through InfraClaude, Claude Code can query Kubernetes clusters, manage Docker containers, check Prometheus metrics, run Terraform plans, and perform security scans — all through natural language.
Architecture
┌─────────────────────────────────────────────────────────┐
│ Claude Code / Claude Desktop │
│ │
│ User: "Check if any pods are crashlooping" │
│ │ │
│ MCP Client │
└─────────────────────────┼────────────────────────────────┘
│ MCP Protocol (stdio)
▼
┌─────────────────────────────────────────────────────────┐
│ InfraClaude MCP Server │
│ │
│ ┌───────────┐ ┌──────────┐ ┌────────────┐ │
│ │ Kubernetes │ │ Docker │ │ Prometheus │ │
│ │ 16 tools │ │ 9 tools │ │ 5 tools │ │
│ └───────────┘ └──────────┘ └────────────┘ │
│ ┌───────────┐ ┌──────────┐ ┌────────────┐ │
│ │ Terraform │ │ Security │ │ System │ │
│ │ 5 tools │ │ 3 tools │ │ 4 tools │ │
│ └───────────┘ └──────────┘ └────────────┘ │
│ │
│ Safety Layer: Command Guard → RBAC → Audit Logger │
└─────────────────────────────────────────────────────────┘
Plus: Claude Code Hooks + Custom Skills
Features
- 42+ infrastructure tools across Kubernetes, Docker, Prometheus, Terraform, Security, and System
- Safety-first design — 4-tier risk classification (Safe → Caution → Dangerous → Blocked)
- Audit logging — every operation logged for compliance
- Claude Code hooks — pre/post tool-use safety checks and automation
- Custom skills — K8s troubleshooting, incident response, Docker debugging, security review
- MCP resources — cluster info, service health, infrastructure summary
- MCP prompts — guided troubleshooting, capacity planning, security audit workflows
Quick Start
# Clone and build
git clone https://github.com/batu-sonmez/infraclaude.git
cd infraclaude
npm install
npm run build
# Add to Claude Code (~/.claude/claude_code_config.json)
{
"mcpServers": {
"infraclaude": {
"command": "node",
"args": ["/path/to/infraclaude/dist/index.js"],
"env": {
"KUBECONFIG": "~/.kube/config",
"PROMETHEUS_URL": "http://localhost:9090"
}
}
}
}
Then in Claude Code:
"Show me all pods in production"
"Why is my pod crashing?"
"Run a security audit on the default namespace"
"What's the CPU usage of my cluster?"
"Scan the nginx:latest image for vulnerabilities"
Demo
Set up a local demo environment with intentionally broken pods:
make demo-setup
Then try the demo scenarios.
Safety Model
InfraClaude classifies every operation by risk level:
| Level | Action | Example |
|---|---|---|
| Safe | Always allowed | k8s_get_pods, prom_instant_query |
| Caution | Allowed with warning | k8s_scale_deployment |
| Dangerous | Requires confirmation | k8s_delete_pod |
| Blocked | Never allowed | k8s_delete_namespace, terraform_apply |
System namespaces (kube-system, kube-public) have additional protections.
See Safety Documentation for details.
Documentation
- Setup Guide
- Tools Reference — all 42+ tools documented
- Safety Model
- Hooks Guide
- Skills Guide
Tech Stack
| Component | Technology |
|---|---|
| MCP Server | TypeScript + @modelcontextprotocol/sdk |
| Kubernetes | @kubernetes/client-node |
| Docker | dockerode |
| Prometheus | Native fetch API |
| Terraform | CLI wrapper |
| Security | Trivy, Gitleaks CLI |
| Testing | Vitest |
| CI/CD | GitHub Actions |
License
MIT — see LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。