Open API MCP Server
Provides AI assistants with access to OpenAPI specifications, enabling API discovery, schema retrieval, and direct API execution with support for OAuth 2.0 and other authentication methods.
README
Open API MCP Server
A Model Context Protocol (MCP) server that provides AI assistants with access to OpenAPI specifications, supporting OAuth 2.0 authentication flows.
Features
- 📖 OpenAPI Support: Load and parse OpenAPI 2.0, 3.0, and 3.1 specifications
- 🔐 OAuth 2.0 Authentication: Full support for Password, Client Credentials, and Authorization Code flows
- 🔑 Multiple Auth Methods: API Key, Bearer Token, Basic Auth, and OAuth 2.0
- 🔍 Smart Search: Search and filter API endpoints by path, method, tags, or keywords
- 📝 Detailed Schemas: Get complete request/response schemas for any endpoint
- 🚀 API Execution: Call APIs directly with automatic authentication
- 📊 Request Logging: Track all API calls with detailed logs
Installation
Option 1: Using npx (Recommended)
No installation required! Use npx to run directly:
npx @yanhao-an/open-api-mcp
Option 2: Global Installation
npm install -g @yanhao-an/open-api-mcp
open-api-mcp
Option 3: From Source (Development)
git clone https://github.com/HaoYan-A/open-api-mcp.git
cd open-api-mcp
npm install
npm run build
npm start
Configuration
Create a .env file in the project root:
# Required: OpenAPI Specification URL
OPENAPI_SPEC_URL=http://127.0.0.1/v3/api-docs
# Authentication Mode
# Options: none, apiKey, bearer, basic, oauth2
AUTH_MODE=oauth2
# OAuth 2.0 Configuration (if AUTH_MODE=oauth2)
OAUTH_FLOW=clientCredentials
OAUTH_CLIENT_ID=your-client-id
OAUTH_CLIENT_SECRET=your-client-secret
OAUTH_TOKEN_URL=https://auth.example.com/oauth/token
OAUTH_SCOPE=read write
# Logging
LOG_LEVEL=info
LOG_REQUESTS=true
LOG_RESPONSES=true
Authentication Modes
1. OAuth 2.0 - Client Credentials Flow
Best for machine-to-machine communication:
AUTH_MODE=oauth2
OAUTH_FLOW=clientCredentials
OAUTH_CLIENT_ID=your-client-id
OAUTH_CLIENT_SECRET=your-client-secret
OAUTH_TOKEN_URL=https://auth.example.com/oauth/token
OAUTH_SCOPE=api.access
2. OAuth 2.0 - Password Flow
For username/password authentication:
AUTH_MODE=oauth2
OAUTH_FLOW=password
OAUTH_CLIENT_ID=your-client-id
OAUTH_CLIENT_SECRET=your-client-secret
OAUTH_USERNAME=user@example.com
OAUTH_PASSWORD=your-password
OAUTH_TOKEN_URL=https://auth.example.com/oauth/token
OAUTH_SCOPE=read write
3. OAuth 2.0 - Authorization Code Flow
For user authorization with redirect:
AUTH_MODE=oauth2
OAUTH_FLOW=authorizationCode
OAUTH_CLIENT_ID=your-client-id
OAUTH_CLIENT_SECRET=your-client-secret
OAUTH_AUTH_URL=https://auth.example.com/oauth/authorize
OAUTH_TOKEN_URL=https://auth.example.com/oauth/token
OAUTH_REDIRECT_URI=http://localhost:3000/callback
OAUTH_CODE=<authorization-code-from-redirect>
4. API Key Authentication
AUTH_MODE=apiKey
API_KEY=your-api-key
API_KEY_HEADER=Authorization
API_KEY_PREFIX=Bearer
5. Bearer Token
AUTH_MODE=bearer
BEARER_TOKEN=your-bearer-token
6. Basic Authentication
AUTH_MODE=basic
BASIC_AUTH_USER=username
BASIC_AUTH_PASS=password
Usage
Running the Server
# Development mode
npm run dev
# Production mode
npm start
# Test with MCP Inspector
npm run inspect
Using with Claude Desktop
Add to your Claude Desktop configuration (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"open-api": {
"command": "npx",
"args": ["-y", "@yanhao-an/open-api-mcp"],
"env": {
"OPENAPI_SPEC_URL": "http://127.0.0.1/v3/api-docs",
"AUTH_MODE": "oauth2",
"OAUTH_FLOW": "clientCredentials",
"OAUTH_CLIENT_ID": "your-client-id",
"OAUTH_CLIENT_SECRET": "your-client-secret",
"OAUTH_TOKEN_URL": "https://auth.example.com/oauth/token"
}
}
}
}
Or if you installed it globally:
{
"mcpServers": {
"open-api": {
"command": "open-api-mcp",
"args": [],
"env": {
"OPENAPI_SPEC_URL": "http://127.0.0.1/v3/api-docs",
"AUTH_MODE": "oauth2",
"OAUTH_FLOW": "clientCredentials",
"OAUTH_CLIENT_ID": "your-client-id",
"OAUTH_CLIENT_SECRET": "your-client-secret",
"OAUTH_TOKEN_URL": "https://auth.example.com/oauth/token"
}
}
}
}
MCP Tools
1. listApi
List all available API endpoints with optional filtering.
Parameters:
tag(optional): Filter by tag namelimit(optional): Maximum results (default: 50)offset(optional): Skip results (default: 0)
Example:
{
"tag": "users",
"limit": 20
}
2. searchApi
Search for API endpoints by keyword.
Parameters:
query(required): Search querymethod(optional): HTTP method filtertag(optional): Tag filter
Example:
{
"query": "user",
"method": "GET"
}
3. getApiSchema
Get detailed schema for a specific endpoint.
Parameters:
path(required): API path (e.g.,/users/{id})method(required): HTTP method
Example:
{
"path": "/api/users/{id}",
"method": "GET"
}
4. callApi
Execute an API request with authentication.
Parameters:
path(required): API pathmethod(required): HTTP methodparams(optional): Query parametersbody(optional): Request bodyheaders(optional): Additional headers
Example:
{
"path": "/api/users",
"method": "POST",
"body": {
"name": "John Doe",
"email": "john@example.com"
}
}
5. getAuthStatus
Get current authentication status and configuration.
6. getApiInfo
Get OpenAPI document information (title, version, base URL, tags).
Project Structure
open-api-mcp/
├── src/
│ ├── auth/ # Authentication management
│ │ ├── auth-manager.ts # Main auth manager
│ │ ├── oauth2-handler.ts # OAuth 2.0 orchestrator
│ │ ├── token-manager.ts # Token storage
│ │ └── flows/ # OAuth 2.0 flow implementations
│ ├── openapi/ # OpenAPI document handling
│ │ ├── loader.ts # Load from URL
│ │ ├── parser.ts # Parse and index
│ │ └── cache.ts # Document caching
│ ├── api/ # API execution
│ │ └── caller.ts # HTTP client with auth
│ ├── tools/ # MCP tools
│ │ └── index.ts # Tool registration
│ ├── logger.ts # Logging utility
│ ├── types.ts # TypeScript types
│ └── index.ts # Main entry point
├── package.json
├── tsconfig.json
└── .env
Development
# Install dependencies
npm install
# Build TypeScript
npm run build
# Watch mode
npm run watch
# Test with inspector
npm run inspect
OAuth 2.0 Flow Details
Token Management
- Automatic Refresh: Tokens are automatically refreshed when expired
- Token Caching: Valid tokens are cached to reduce auth requests
- Expiry Buffer: Tokens are refreshed 60 seconds before expiry
Flow Selection Guide
- Client Credentials: Best for server-to-server, no user context
- Password Flow: For trusted applications with username/password
- Authorization Code: For user-authorized access with redirect flow
Logging
The server provides detailed logging for debugging:
# Log levels: debug, info, warn, error
LOG_LEVEL=debug
# Enable/disable request logging
LOG_REQUESTS=true
# Enable/disable response logging
LOG_RESPONSES=true
Request logs include:
- Timestamp
- HTTP method and URL
- Authentication mode
- Request parameters and body
- Response status and data
- Request duration
Troubleshooting
"Failed to load OpenAPI document"
- Verify
OPENAPI_SPEC_URLis correct and accessible - Check network connectivity
- Ensure the URL returns valid OpenAPI JSON/YAML
"OAuth 2.0 flow failed"
- Verify OAuth credentials (client ID, secret)
- Check token URL is correct
- Ensure required scopes are available
- Review server logs for specific error messages
"Token refresh failed"
- Verify refresh token is valid
- Check if refresh tokens are supported by the auth server
- Try re-authenticating to get a new token
License
MIT
Contributing
Contributions are welcome! Please feel free to submit issues or pull requests.
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。