CDS Components MCP Server
Enables AI assistants to search, understand, and generate code for design system components by syncing and indexing a component library.
README
CDS Components MCP Server
An MCP (Model Context Protocol) server for your component library and design system, enabling AI assistants to search, understand, and generate code using your components.
Features
- Component Search: Search by name, props, or functionality
- Component Details: Retrieve full documentation including props, examples, and dependencies
- Similar Components: Find alternatives based on requirements
- Prop Validation: Check if props match library conventions
- Code Generation: Generate boilerplate code with correct imports and props
Architecture
src/
├── types.ts # TypeScript interfaces
├── config.ts # Configuration management
├── repo-sync.ts # Git repository synchronization
├── parsers/
│ ├── component.ts # Parse React/TypeScript components
│ ├── story.ts # Parse Storybook stories
│ ├── tokens.ts # Parse design tokens
│ └── mdx.ts # Parse MDX documentation
├── indexer.ts # Orchestrates all parsers
├── search.ts # Search and similarity algorithms
└── index.ts # MCP server entry point
Installation
# Install dependencies
pnpm install
# Build the project
pnpm build
Configuration
Create a .env file from the example:
cp .env.example .env
Configure your repository settings:
REPO_OWNER=your-org
REPO_NAME=your-design-system-repo
REPO_BRANCH=main
PACKAGES_PATH=packages/components,packages/tokens
Usage
Running the Server
node dist/index.js
Configuring with Claude Desktop
Add to your Claude Desktop config (~/.config/claude/claude_desktop_config.json):
{
"mcpServers": {
"cds-components": {
"command": "node",
"args": ["/path/to/cds-components-mcp/dist/index.js"],
"env": {
"REPO_OWNER": "your-org",
"REPO_NAME": "your-design-system",
"REPO_BRANCH": "main",
"PACKAGES_PATH": "packages/components,packages/tokens"
}
}
}
}
Available Tools
1. search_components
Search for components by name, props, or functionality.
Arguments:
query(string, required): Search queryfilterBy(string, optional): Filter by "name", "props", "functionality", or "all"
Example:
Find all button components
2. get_component_details
Retrieve full documentation for a specific component.
Arguments:
componentName(string, required): Name of the component
Example:
Get details for Button component
3. find_similar_components
Suggest alternative components based on requirements.
Arguments:
requirements(string, required): Description of component requirementscomponentType(string, optional): Type or category filter
Example:
Find components for displaying user profiles
4. validate_component_usage
Check if props and patterns match library conventions.
Arguments:
componentName(string, required): Component to validateprops(object, required): Props object to validate
Example:
Validate Button usage with props {variant: "primary", size: "large"}
5. generate_component_code
Generate boilerplate code with correct imports and props.
Arguments:
componentName(string, required): Component nameprops(array, optional): List of prop namesincludeStyles(boolean, optional): Include style templates
Example:
Generate code for Button component
How It Works
- Repository Sync: Clones or pulls your design system repository
- Indexing: Parses all components, stories, tokens, and docs using TypeScript AST
- Search: Uses relevance scoring to find matching components
- Tool Execution: Returns formatted documentation and code snippets
Supported File Types
- Components:
.tsx,.jsx - Stories:
.stories.tsx,.stories.ts - Tokens:
.json,.ts,.js,.css,.scss - Documentation:
.mdx,.md
Development
# Watch mode for development
pnpm watch
# Build
pnpm build
License
ISC
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。