VULK MCP Server
Enables AI assistants to generate, edit, and deploy production-ready full-stack applications through VULK's automated development pipeline. It supports multi-file code generation, browser verification, and exports to web, PWA, and mobile platforms.
README
<p align="center"> <img src="https://vulk.dev/images/vulk-icon.svg" alt="VULK" width="60" height="60" /> </p>
<h1 align="center">VULK MCP Server</h1>
<p align="center"> <strong>Build, deploy, and export full-stack applications from any AI assistant.</strong> </p>
<p align="center"> <a href="https://www.npmjs.com/package/vulk-mcp-server"><img src="https://img.shields.io/npm/v/vulk-mcp-server?color=0D9373" alt="npm" /></a> <a href="https://vulk.dev"><img src="https://img.shields.io/badge/vulk.dev-live-0D9373" alt="VULK" /></a> <a href="https://github.com/vulkdev/vulk-mcp-server/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-MIT-blue" alt="MIT License" /></a> </p>
<p align="center"> Give Claude, Cursor, Windsurf, VS Code Copilot, or Gemini CLI the ability to generate, edit, deploy, and export production-ready applications — powered by <a href="https://vulk.dev">VULK</a>. </p>
What This Does
This MCP server connects AI coding assistants to VULK's full application platform. Say "build me a project management dashboard" and it triggers real AI generation — not templates, not scaffolding, but a complete application.
Generation
- 16+ LLM models — Claude Opus 4.6, GPT-5.4, Gemini 3.1 Pro, DeepSeek V3, Grok 4, and more
- Full-stack output — React + Tailwind + routing + API endpoints + database schemas
- Real-time streaming — watch files being generated in your terminal
- Auto-fix pipeline — linting, browser verification, self-debugging loop
Export & Deploy
- Web — Deploy to Cloudflare Pages with custom domains + SSL
- PWA — Progressive Web App with offline support
- APK — Android app builds (server-side compilation)
- iOS — iOS app builds via Flutter
- Flutter — Web-to-Flutter AI converter for native mobile
- Custom domains — Automatic SSL, DNS configuration included
Platform
- Bring Your Own Model (BYOM) — use your own API keys for any provider
- Figma import — convert Figma designs to code
- GitHub integration — push generated code directly
- Team collaboration — shared workspaces, role-based permissions
- 8 languages — EN, PT, FR, DE, ES, IT, JA, HI
Quick Setup
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"vulk": {
"command": "npx",
"args": ["-y", "vulk-mcp-server"],
"env": {
"VULK_API_KEY": "vk_sk_your_key_here"
}
}
}
}
Cursor
Settings → MCP Servers → Add:
{
"vulk": {
"command": "npx",
"args": ["-y", "vulk-mcp-server"],
"env": {
"VULK_API_KEY": "vk_sk_your_key_here"
}
}
}
VS Code (GitHub Copilot)
Create .vscode/mcp.json:
{
"servers": {
"vulk": {
"command": "npx",
"args": ["-y", "vulk-mcp-server"],
"env": {
"VULK_API_KEY": "vk_sk_your_key_here"
}
}
}
}
Windsurf
Add to MCP settings:
{
"vulk": {
"command": "npx",
"args": ["-y", "vulk-mcp-server"],
"env": {
"VULK_API_KEY": "vk_sk_your_key_here"
}
}
}
Gemini CLI
Install the extension directly from GitHub:
gemini extensions install VULK-dev/vulk-mcp-server
Then set your API key:
export VULK_API_KEY="vk_sk_your_key_here"
Or configure it in the extension settings when prompted. The extension uses the same MCP server under the hood, giving you access to all VULK tools (generate, edit, deploy, etc.) directly from Gemini CLI.
Get Your API Key
- Go to vulk.dev/settings/api-keys
- Click Create API Key
- Copy the key (starts with
vk_sk_)
Free accounts get 3 generations/month. Upgrade for more.
Tools
generate — Build a new app
"Build a modern SaaS dashboard with user auth, analytics charts, team management, and dark mode"
Creates a project, triggers AI generation, and returns all generated files with a live preview URL. Generation runs through VULK's full pipeline — intent analysis, multi-file code generation, auto-fixing, browser verification, and quality scoring.
edit — Modify an existing project
"Add a settings page with tabs for Profile, Billing, and Notifications"
Sends your instruction to VULK's AI with full context of all existing project files. The AI decides which files to create or modify.
list — See your projects
Returns all your VULK projects with IDs, prompts, dates, and deployment URLs.
get — Project details
Get status, metadata, and URLs for a specific project.
files — Read source code
Download every file from a project — paths, content, language detection.
deploy — Ship to production
Deploy to Cloudflare Pages and get a live production URL with custom domain support.
models — Available LLM models
List all available models on your plan — Claude Opus 4.6, Sonnet 4.6, GPT-5.4, Gemini 3.1 Pro, DeepSeek V3, Grok 4, and more.
usage — Check your limits
View API request counts, credits remaining, and rate limit status.
subscribe — Upgrade your plan
Get a checkout link to upgrade. Plans from $19.99/mo to $299/mo.
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
VULK_API_KEY |
Yes | — | Your VULK API key (vk_sk_...) |
VULK_API_BASE |
No | https://vulk.dev |
API base URL |
How It Works
You → "Build me a task manager"
↓
MCP Server → POST /api/v1/projects (create record)
↓
MCP Server → POST /api/agent/stream (trigger AI generation)
↓
VULK Agent → Intent analysis → Code generation → Auto-fix → Browser verify
↓
MCP Server ← SSE stream (file_start, file_delta, file_complete events)
↓
You ← { files: [...], previewUrl, editorUrl }
The generation pipeline includes:
- Intent analysis — understands what kind of app you want
- ReAct agent loop — plans and generates files with tool use
- Auto-fixer — deterministic code fixes (imports, routing, styling)
- Browser verification — renders the app, catches errors, self-heals
- Quality scoring — ensures the output meets production standards
Pricing
| Plan | Price | Credits/month | LLM Models | Best For |
|---|---|---|---|---|
| Free | $0 | 3 generations | Basic | Trying it out |
| Builder | $19.99/mo | 1,000 | Basic (Haiku, Flash, Mini) | Getting started |
| Pro | $39.99/mo | 2,500 | All 16+ models | Power users |
| Team | $79.99/mo | 5,000 | All + team collaboration | Small teams |
| Max | $199/mo | 10,000 | All + BYOM + white-label | Agencies |
| Business | $299/mo | 20,000 | Everything + SSO + SLA | Organizations |
Credits are token-based — simple apps use ~100 credits, complex ones ~500+. Full pricing details.
Development
git clone https://github.com/vulkdev/vulk-mcp-server.git
cd vulk-mcp-server
npm install
npm run build
VULK_API_KEY=vk_sk_... node dist/index.js
Links
- VULK — AI-powered application builder
- API Keys — Get your key
- Pricing — Plans and pricing
- Documentation — Full docs
- Compare — VULK vs Bolt vs Lovable vs v0
License
MIT
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。