OpenShift MCP Server
Provides AI assistants with comprehensive OpenShift/Kubernetes cluster management capabilities through the oc CLI.
README
OpenShift MCP Server
A Model Context Protocol (MCP) server that provides AI assistants with comprehensive OpenShift/Kubernetes cluster management capabilities through the oc command-line interface.
Demo Video
Watch the oc-new-app tool deploy a complete Spring Boot application from GitHub to OpenShift:
https://github.com/user-attachments/assets/f4bace3f-755b-462f-8a8a-680e4dc02129
Features
- Cluster Information: Access comprehensive cluster status, nodes and configuration via MCP resources
- Resource Description: Describe any resource with multiple output formats including human-readable summaries
- Complete Resource Management: Create, read, update, delete, and patch all OpenShift/Kubernetes resources
- Application Deployment: Deploy applications from Git repositories with S2I builds and automatic route creation
- Operator Management: Install operators via OLM, Helm, or direct manifests
- Cluster Operations: Scaling, monitoring, troubleshooting and management
- Build Operations: Start and monitor OpenShift builds
- Scaling Operations: Scale deployments, replicasets and statefulsets
- Logging: Retrieve logs from pods and builds
- Multi-Transport Support: STDIO and HTTP/SSE transports for different integration scenarios
- Comprehensive Testing: 99 unit tests ensuring production reliability
- Rich Error Handling: Detailed troubleshooting guidance and actionable error messages
- Troubleshooting Prompts: Interactive troubleshooting guides for common OpenShift scenarios
- Log Sampling: Sample and analyze pod logs with intelligent pattern detection and context
Prerequisites
- Node.js 18+
- OpenShift CLI (
oc) installed and configured - Access to an OpenShift cluster
Installation
git clone https://github.com/sanjaypsachdev/mcp-server-openshift.git
cd mcp-server-openshift
npm install
npm run build
Configuration
Add to your MCP client configuration:
Claude Desktop / Cursor
{
"mcpServers": {
"openshift": {
"command": "node",
"args": ["/path/to/mcp-server-openshift/dist/index.js"],
"env": {
"OPENSHIFT_CONTEXT": "your-context-name",
"OPENSHIFT_NAMESPACE": "your-default-namespace"
}
}
}
}
HTTP/SSE Transport (Remote Access)
# Start HTTP server
npm run start:http
# Connect via MCP remote
npx -y mcp-remote http://localhost:3000/sse --transport sse-only
Tools
Authentication & Access
oc_login- Securely log into OpenShift clusters using token or username/password authentication
API Discovery & Documentation
oc_api_resources- List all available API resources in the cluster with categorizationoc_explain- Explain resource schemas, fields, and API documentation
Core Resource Management
oc_get- Get OpenShift resources (pods, deployments, services, routes, etc.)oc_create- Create OpenShift resources from manifests or templatesoc_apply- Apply YAML manifests with validation and conflict resolutionoc_delete- Delete resources with safety checks and confirmation optionsoc_patch- Patch resources using strategic merge, JSON merge, or JSON patch operationsoc_describe- Describe resources with multiple output formats
Application Lifecycle
oc_new_app- Deploy applications from Git repositories with S2I buildsoc_scale- Scale deployments, deploymentconfigs, replicasets, and statefulsetsoc_logs- Get logs from pods, deployments, builds with filtering options
Advanced Operations
oc_install_operator- Install operators via OLM, Helm, or direct manifests
Resources
MCP Resources provide read-only access to cluster information:
openshift://cluster-info- Comprehensive cluster status, nodes, namespaces, and eventsopenshift://project-list- Detailed project information with quotas and usage statisticsopenshift://app-templates- Application deployment templates and patterns
Prompts
Interactive troubleshooting and operational guidance:
troubleshoot-openshift-prompt- Comprehensive OpenShift troubleshooting guide for all resource types and cluster issuesmonitoring-prompts- Monitoring and observability guidance for different scenarios
Sampling
Intelligent log analysis and pattern detection:
- Pod Logs Sampling - Automatic log sampling with error pattern detection and context analysis
Usage Examples
Login to Cluster
# Login with token (recommended)
oc_login with server: "https://api.cluster.example.com:6443",
authMethod: "token",
token: "sha256~your-token-here"
# Login with username/password
oc_login with server: "https://api.cluster.example.com:6443",
authMethod: "password",
username: "developer",
password: "your-password"
Discover API Resources
# List all available API resources
oc_api_resources
# List resources for specific API group
oc_api_resources with apiGroup: "apps"
# List only namespaced resources
oc_api_resources with namespaced: true
# Explain a resource schema
oc_explain with resource: "deployment"
# Explain specific field
oc_explain with resource: "pod", field: "spec.containers"
Deploy Application
# Deploy Node.js app from GitHub
oc_new_app with gitRepo: "https://github.com/sclorg/nodejs-ex.git"
Scale Application
# Scale deployment to 3 replicas
oc_scale with name: "my-app", replicas: 3
Patch Resource
# Update deployment labels
oc_patch with resourceType: "deployment", name: "my-app",
patch: '{"metadata":{"labels":{"environment":"production"}}}'
Troubleshoot Issues
# Get pod troubleshooting guidance
Use prompt: troubleshoot-openshift-prompt
Arguments: issueType: "pod", resourceName: "my-app-12345", namespace: "my-project"
# Get deployment troubleshooting guidance
Use prompt: troubleshoot-openshift-prompt
Arguments: issueType: "deployment", resourceName: "my-app", namespace: "my-project"
# Get general cluster troubleshooting guidance
Use prompt: troubleshoot-openshift-prompt
Arguments: issueType: "cluster", symptoms: "nodes not ready"
Development
# Setup
npm install
# Build
npm run build
# Test
npm test
# Start (STDIO)
npm start
# Start (HTTP)
npm run start:http
# Development mode
npm run dev
Transport Modes
STDIO (Default)
- Direct MCP client integration
- Lower latency
- Recommended for local development
HTTP/SSE
- Remote access capability
- Web integration friendly
- Container deployment ready
Architecture
src/
├── index.ts # Main server entry point
├── tools/ # Tool implementations
├── resources/ # MCP resources
├── prompts/ # Interactive prompts
├── sampling/ # Log sampling and analysis
├── models/ # Zod validation schemas
└── utils/ # OpenShift CLI wrapper
Security
Authentication Security
- Token Authentication: Preferred method for automation and production use
- Password Authentication: Available but token authentication is recommended
- HTTPS Enforcement: All cluster connections must use HTTPS
- URL Validation: Server URLs validated to prevent SSRF attacks
- Private IP Blocking: Prevents connections to internal/metadata services
Operational Security
- RBAC Compliance: Respects OpenShift RBAC permissions
- No Credential Storage: Credentials are not stored or transmitted by the server
- User Permissions: Executes with the same permissions as the authenticated user
- Input Validation: Comprehensive validation of all inputs and parameters
- Secure Defaults: Conservative security settings by default
Best Practices
- Use Service Account Tokens: For automation and CI/CD pipelines
- Regular Token Rotation: Rotate authentication tokens regularly
- TLS Certificate Validation: Always validate TLS certificates in production
- Least Privilege: Use accounts with minimal required permissions
- Session Management: Use
oc logoutto clear credentials when done
License
MIT License - see LICENSE file for details.
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。
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。
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。
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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。