VengeanceUI MCP Server
Enables AI assistants to browse, search, and retrieve source code of 129 VengeanceUI components from their GitHub repository.
README
VengeanceUI MCP Server
An Model Context Protocol (MCP) server that brings 129 beautiful VengeanceUI components directly to your AI assistant! Browse, search, and retrieve component source code from the VengeanceUI GitHub repository seamlessly within Claude Desktop, Cursor, or any MCP-enabled tool.
✨ Features
- 🎨 129+ Components: Access the complete VengeanceUI component library
- 🔍 Smart Search: Search by name, category, tags, or description with fuzzy matching
- 📦 Full Source Code: Get complete component code with metadata
- 🗂️ Category Filtering: Browse by Buttons, Cards, Forms, Navigation, and more
- 📊 Component Metadata: View dependencies, tags, file size, and source URLs
- ⚡ Smart Caching: Built-in 5-minute cache for faster responses
- 🔄 Real-time Sync: Components loaded directly from the latest GitHub repository
🚀 Quick Start
Prerequisites
- Node.js: v18 or higher (tested with v22.17.0)
- npm: Latest version
- Claude Desktop or Cursor IDE (or any MCP-compatible client)
Installation
-
Clone this repository:
git clone https://github.com/krockxz/VengeanceUI-MCP.git cd VengeanceUI-MCP -
Install dependencies:
npm install -
Build the project:
npm run buildYou should see:
> vengeanceui-mcp-server@1.0.0 build > tsc -
Verify the build:
ls -lh dist/server.js
⚙️ Configuration
For Claude Desktop
-
Locate your configuration file:
- Linux:
~/.config/Claude/claude_desktop_config.json - macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
- Linux:
-
Add the VengeanceUI server configuration:
{ "mcpServers": { "vengeanceui": { "command": "node", "args": ["/absolute/path/to/mcp VengeanceUI/dist/server.js"] } } }⚠️ Important: Replace
/absolute/path/to/with your actual absolute path!Example on Linux:
"args": ["/home/kunal/Documents/mcp VengeanceUI/dist/server.js"] -
Restart Claude Desktop:
- Linux: Press
Ctrl + Ror fully restart the application - macOS/Windows: Fully quit and restart Claude Desktop
- Linux: Press
-
Verify the server is running:
- Look for the 🔨 tool icon in the bottom-right corner of Claude Desktop
- Click it to see available MCP servers
- You should see "VengeanceUI MCP Server" listed
For Cursor IDE
-
Create or edit
.cursor/mcp-servers.jsonin your project:{ "mcpServers": { "vengeanceui": { "command": "node", "args": ["/absolute/path/to/mcp VengeanceUI/dist/server.js"] } } } -
Restart Cursor IDE
Optional: GitHub Token for Higher Rate Limits
Adding a GitHub token increases API rate limits from 60 to 5000 requests/hour.
-
Create a GitHub Personal Access Token:
- Go to https://github.com/settings/tokens
- Click "Generate new token (classic)"
- Select scope:
public_repo(read access to public repositories) - Copy the generated token (starts with
ghp_)
-
Update your configuration:
{ "mcpServers": { "vengeanceui": { "command": "node", "args": ["/absolute/path/to/mcp VengeanceUI/dist/server.js"], "env": { "GITHUB_TOKEN": "ghp_your_token_here" } } } } -
Restart your MCP client
🛠️ Available Tools
The server provides 7 powerful tools for working with VengeanceUI components:
1. list_components
List all available VengeanceUI components with their categories and descriptions.
Parameters:
category(optional): Filter by category (e.g., "Buttons", "Cards", "Forms")limit(optional): Maximum number of components to return
Example prompts:
- "List all VengeanceUI components"
- "Show me the first 20 components"
- "List all components in the Buttons category"
2. search_components
Search for components using fuzzy matching across names, categories, tags, and descriptions.
Parameters:
query(required): Search termlimit(optional): Maximum results to return (default: 10)
Example prompts:
- "Search for button components"
- "Find animated card components"
- "Search for gradient effects"
3. get_component_code
Retrieve the complete source code for a specific component with metadata.
Parameters:
component_name(required): Exact or partial component nameinclude_metadata(optional): Include metadata like tags, dependencies (default: true)
Example prompts:
- "Get the code for AnimatedButton"
- "Show me the source code for GradientCard"
- "Get button.tsx code"
4. get_components_by_category
Get all components in a specific category.
Parameters:
category(required): Component category name
Example prompts:
- "Show all components in the Forms category"
- "What's in the Navigation category?"
5. get_component_info
Get detailed metadata about a component including dependencies, file size, and tags.
Parameters:
component_name(required): Component name
Example prompts:
- "Get info for AnimatedButton"
- "What are the dependencies for GradientCard?"
6. list_categories
List all component categories with component counts.
Example prompts:
- "List all categories"
- "What categories are available in VengeanceUI?"
7. refresh_cache
Force refresh the component cache from the GitHub repository.
Example prompts:
- "Refresh the VengeanceUI component cache"
- "Update the component list"
💡 Usage Examples
Once configured in Claude Desktop, try these natural language prompts:
Discovery
"What VengeanceUI components are available?"
"List all categories in VengeanceUI"
"How many components are there in total?"
Search
"Find all button components"
"Search for gradient or animated components"
"Show me navigation components"
Get Code
"Get the code for button.tsx"
"Show me the AnimatedButton component"
"I need the source code for GradientCard"
Browse by Category
"Show all components in the Buttons category"
"What forms components are available?"
"List all navigation components"
Component Details
"Get detailed info for AnimatedButton"
"What dependencies does the GradientCard component have?"
"Show me the metadata for button.tsx"
🔧 Development
Development Mode
Run the server in watch mode:
npm run dev
Build
Compile TypeScript to JavaScript:
npm run build
Watch Mode
Automatically rebuild on file changes:
npm run watch
📖 How It Works
-
Component Discovery: The server connects to the VengeanceUI GitHub repository and scans the following directories:
registry/new-york/- Main component registrysrc/components/- Source components
-
Intelligent Parsing:
- Extracts component names from filenames
- Parses code to detect categories, tags, and dependencies
- Analyzes JSDoc comments for descriptions
- Identifies animation, styling, and interaction patterns
-
Smart Caching:
- Components are cached for 5 minutes
- Reduces GitHub API calls
- Automatic refresh on cache expiration
-
MCP Integration:
- Implements the Model Context Protocol
- Exposes 7 tools via stdio transport
- Provides structured responses with metadata
🐛 Troubleshooting
Server doesn't appear in Claude Desktop
-
Check configuration file syntax:
cat ~/.config/Claude/claude_desktop_config.json | jq .If
jqshows an error, your JSON is malformed. -
Verify the server file exists:
ls -lh "/absolute/path/to/mcp VengeanceUI/dist/server.js" -
Check Claude Desktop logs:
# Find log directory find ~ -name "*Claude*" -type d 2>/dev/null | grep -i log # View logs tail -f ~/.config/Claude/logs/mcp*.log -
Test the server manually:
cd "mcp VengeanceUI" node dist/server.js
No components are loading
-
Check GitHub API rate limits:
- Without token: Limited to 60 requests/hour
- Solution: Add a GitHub token (see Configuration section)
-
Force refresh the cache:
- Use the prompt: "Refresh the VengeanceUI component cache"
Components are outdated
The cache refreshes every 5 minutes. To force an immediate refresh, use the refresh_cache tool via this prompt:
"Refresh the VengeanceUI component cache"
🙏 Acknowledgments
- VengeanceUI - For the incredible component library featuring beautiful, animated components
- Model Context Protocol - For the server framework enabling AI assistant integration
- shadcn-ui-mcp-server - Inspiration for this project
🔗 Links
- VengeanceUI Repository: https://github.com/Ashutoshx7/VengeanceUI
- MCP Documentation: https://modelcontextprotocol.io
- Issues & Support: Open an issue in this repository
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。