MCP Startup Framework
A framework for building MCP servers on Cloudflare Workers with OAuth, PostgreSQL, and Stripe.
MCP Startup Framework
A complete framework for building remote Model Context Protocol (MCP) servers on Cloudflare Workers with OAuth authentication, PostgreSQL database, and Stripe-powered paid tools.
🚀 Features
- 🔐 OAuth 2.1 Provider - Complete authentication with user registration/login
- 🗄️ PostgreSQL Integration - Scalable production-ready database with connection pooling
- 💳 Paid Tools Framework - Stripe-powered premium tools with subscription support
- 🌐 Platform Agnostic - Optimized for Cloudflare Workers but deployable anywhere (Vercel, Railway, AWS, self-hosted)
- 🛠️ MCP Tools Framework - Modular tool system with user context
- 🔌 REST API Routes - Easy-to-use system for adding custom endpoints
- 🎨 Custom Views - Extensible UI system for privacy policies, ToS, documentation pages
- 📡 Streamable HTTP Transport - Works with
mcp-remote
for local Claude Desktop integration - 📱 Cross-Platform - Works on desktop, web, and mobile MCP clients
⚡ Quick Start
Prerequisites
- Node.js 18+
- Cloudflare account
- PostgreSQL database (see database setup)
- Stripe account (for paid tools)
Installation
# Clone the repository
git clone https://github.com/f/mcp-startup-framework
cd mcp-startup-framework
# Install dependencies
npm install
# Copy environment template
cp .dev.vars.example .dev.vars
Configuration
Edit .dev.vars
with your settings:
DATABASE_URL="postgresql://username:password@host:port/database"
JWT_SECRET="your-super-secret-jwt-key"
COOKIE_ENCRYPTION_KEY="32-character-encryption-key"
STRIPE_SECRET_KEY="sk_test_your_stripe_key"
STRIPE_PRICE_ID_FOR_PREMIUM_MATH="price_your_stripe_price_id"
BASE_URL="https://your-domain.workers.dev"
Development
# Start development server
npm run dev
# Initialize database (in another terminal)
curl -X POST http://localhost:8787/init-db
# Visit http://localhost:8787 to register/login
🛠️ Adding Tools & Routes
Everything starts from src/index.ts
:
// Register MCP tools
backend
.registerTool(registerAddTool)
.registerTool(registerGreetingTool)
.registerTool(registerPremiumMathTool);
// Add REST API routes
backend
.route('GET', '/api/status', (c) => c.json({ status: 'ok' }))
.authRoute('GET', '/api/profile', (c, userContext) =>
c.json({ user: userContext })
);
See Tools Guide and Routes Guide for details.
Custom Views
Extend beyond MCP with custom pages:
// Add privacy policy, terms of service, documentation
backend
.route('GET', '/privacy', (c) => c.html(generatePrivacyPage()))
.route('GET', '/terms', (c) => c.html(generateTermsPage()))
.route('GET', '/docs', (c) => c.html(generateDocsPage()));
See Views Guide for creating custom pages.
🚀 Deployment
# Set production secrets
wrangler secret put JWT_SECRET
wrangler secret put COOKIE_ENCRYPTION_KEY
wrangler secret put STRIPE_SECRET_KEY
# Set custom price IDs per paid tool
wrangler secret put STRIPE_PRICE_ID_FOR_PREMIUM_MATH
# Deploy to Cloudflare
npm run deploy
See Deployment Guide for full instructions.
🔗 Integration Options
Using claude.ai (Pro Plan)
For claude.ai Pro users, you can integrate directly:
- Navigate to claude.ai settings
- Click "Add Integration"
- Enter your server URL:
https://your-domain.workers.dev/sse
- Complete the OAuth flow
Using Claude Desktop
For local Claude Desktop app:
1. Install MCP Remote Client
npm install -g mcp-remote
2. Configure Claude Desktop
Open Claude Desktop settings: Settings > Developer > Edit Config
Add to your claude_desktop_config.json
:
{
"mcpServers": {
"my-startup": {
"command": "npx",
"args": [
"-y",
"mcp-remote",
"https://your-domain.workers.dev/mcp"
]
}
}
}
3. Restart Claude Desktop
After saving, restart Claude Desktop. Your MCP tools will appear in the chat interface.
Transport Options
The framework supports multiple transport protocols:
- Streamable HTTP (default) -
/mcp
endpoint - Server-Sent Events (SSE) -
/sse
endpoint
Both work seamlessly with mcp-remote
for local development and production use.
🌐 Platform Flexibility
While optimized for Cloudflare Workers, this framework runs anywhere:
- Vercel: Deploy with
@vercel/node
runtime - Railway: One-click deploy with built-in PostgreSQL
- AWS Lambda: Use with RDS or Aurora Serverless
- Google Cloud Run: Deploy with Cloud SQL
- Self-hosted: Run with Docker on any VPS
Simply update your database connection and deployment configuration!
Production Database
The framework works with any PostgreSQL provider:
- Neon - Serverless PostgreSQL with autoscaling
- Supabase - PostgreSQL with real-time features
- Railway - Managed PostgreSQL with automatic backups
- Amazon RDS - Enterprise-grade PostgreSQL
- Cloudflare Hyperdrive - Connection pooling for optimal performance
All providers support the scalability needed for production MCP servers.
📚 Documentation
- Setup Guide - Detailed installation and configuration
- Database Setup - PostgreSQL options and optimization
- Development Guide - Project structure and workflows
- Adding Tools - Create free and paid MCP tools
- Custom Routes - Add REST API endpoints
- Views Customization - Customize all UI pages (login, consent, payment, etc.)
- Deployment - Production deployment instructions
🤝 Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
📄 License
MIT License - see LICENSE file for details.
Related Servers
Jinni
A tool to provide Large Language Models with project context by intelligently filtering and concatenating relevant files.
Osmosis Agent Toolkit
An MCP server for the Osmosis Agent Toolkit, enabling interactions with the Osmosis blockchain.
UseGrant MCP Server
Interact with the UseGrant API for programmatic access control and permissions management.
Jetty.io
Work on dataset metadata with MLCommons Croissant validation and creation.
Apache SkyWalking MCP
An MCP server for integrating AI agents with the SkyWalking observability platform and its ecosystem.
BoostSecurity
BoostSecurity MCP acts as a safeguard preventing agents from adding vulnerable packages into projects. It analyzes every package an AI agent introduces, flags unsafe dependencies, and recommends secure, maintained alternatives to keep projects protected.
CAD-MCP
Control CAD software with natural language instructions to perform drawing operations.
Geo Location Demo
Retrieves user geolocation information using EdgeOne Pages Functions and integrates it with large language models via MCP.
MCP Remote Machine Control
Provides remote machine control capabilities, eliminating SSH overhead for token-efficient system operations.
MCP-RAGNAR
A local MCP server implementing Retrieval-Augmented Generation (RAG) with sentence window retrieval and support for multiple file types.