Skip to main content
Glama
README.mdβ€’9.45 kB
# ACF MCP-Proxy Deployment Guide Deploy your Agentic Control Framework (ACF) to the cloud in under 1 hour with zero code changes using mcp-proxy bridge. ## πŸš€ Quick Start (Commercial Ready) The fastest path to commercial deployment using mcp-proxy to convert your existing STDIO-based ACF server to HTTP/SSE: ```bash # Deploy with authentication and monetization ./quick-deploy.sh gcp --with-auth # Deploy proxy only (no auth) ./quick-deploy.sh railway --proxy-only # Test locally first ./quick-deploy.sh local --with-auth ``` ## πŸ—οΈ Architecture ``` β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” HTTP/SSE β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” STDIO β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ Claude/ β”‚ ◄─────────────► β”‚ Auth Proxy β”‚ ◄──────────► β”‚ mcp-proxy β”‚ ◄─► β”‚ Your ACF β”‚ β”‚ Client β”‚ β”‚ (Optional) β”‚ β”‚ (Bridge) β”‚ β”‚ Server β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β–² β–² β”‚ Deploy to β”‚ β–Ό β–Ό β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”Œβ”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β” β”‚ Cloud β”‚ β”‚ Cloud β”‚ β”‚ Platform β”‚ β”‚ Platform β”‚ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜ ``` ## 🎯 Features ### Core Features - **Zero Code Changes**: Use your existing ACF server as-is - **Full Functionality**: All 80+ tools work immediately - **HTTP/SSE Bridge**: Convert STDIO to web-compatible protocols - **Production Ready**: Battle-tested mcp-proxy foundation ### Commercial Features (with `--with-auth`) - **User Authentication**: Token-based access control - **Tiered Pricing**: Free, Pro, Enterprise tiers - **Stripe Integration**: Automated payment processing - **Usage Tracking**: Per-user request monitoring - **Rate Limiting**: Prevent abuse and manage costs - **Professional UI**: Landing page and user dashboard ## πŸ“¦ Supported Platforms | Platform | Deployment Time | Cost | Best For | |----------|----------------|------|----------| | **Google Cloud Run** | 5 minutes | Pay-per-request | Production, Auto-scaling | | **Railway** | 2 minutes | $5/month | Simple deployment | | **Fly.io** | 3 minutes | Free tier available | Edge performance | | **Local Docker** | 1 minute | Free | Development, Testing | ## πŸ”§ Prerequisites - Node.js 22+ (LTS) and npm - Docker and Docker Compose - Platform-specific CLI (gcloud, railway, or fly) ## πŸš€ Deployment Options ### Option 1: Google Cloud Run (Recommended) Best for production with automatic scaling and pay-per-request pricing. ```bash # Set your project ID export GCP_PROJECT_ID="your-project-id" # Deploy with full commercial features ./quick-deploy.sh gcp --with-auth # Or deploy proxy only ./quick-deploy.sh gcp --proxy-only ``` **Requirements:** - Google Cloud account with billing enabled - `gcloud` CLI installed and authenticated ### Option 2: Railway (Simplest) Perfect for quick testing and simple deployments. ```bash # Deploy to Railway ./quick-deploy.sh railway --with-auth ``` **Requirements:** - Railway account (free tier available) - `railway` CLI will be installed automatically ### Option 3: Fly.io (Performance) Great for edge performance with global distribution. ```bash # Deploy to Fly.io ./quick-deploy.sh fly --with-auth ``` **Requirements:** - Fly.io account - `fly` CLI will be installed automatically ### Option 4: Local Development Test your deployment locally before going to production. ```bash # Start local deployment with Docker Compose ./quick-deploy.sh local --with-auth # Test proxy only ./quick-deploy.sh test ``` ## πŸ’° Monetization Setup When deploying with `--with-auth`, you get a complete commercial solution: ### 1. Configure Payment Processing Set up Stripe for payment processing: ```bash export STRIPE_SECRET_KEY="sk_test_..." export STRIPE_PUBLISHABLE_KEY="pk_test_..." export STRIPE_PRO_PRICE_ID="price_..." export STRIPE_ENTERPRISE_PRICE_ID="price_..." ``` ### 2. Set Up User Database Configure Supabase for user management: ```bash export SUPABASE_URL="https://your-project.supabase.co" export SUPABASE_ANON_KEY="eyJ..." ``` ### 3. Pricing Tiers | Tier | Price | Requests/Month | Tools | Support | |------|-------|----------------|-------|---------| | **Free** | $0 | 100 | 15 basic tools | Community | | **Pro** | $29 | 10,000 | All 80+ tools | Priority | | **Enterprise** | $199 | 100,000 | All + Custom | 24/7 | ### 4. Landing Page Professional landing page included at your deployed URL with: - Feature showcase - Pricing tiers - User registration - Payment processing - Token management ## πŸ” Authentication & Security ### Token Management The auth proxy handles token validation: ```javascript // Free tier tokens acf_free_1703123456_abc123 // Pro tier tokens acf_pro_1703123456_def456 // Enterprise tokens acf_enterprise_1703123456_ghi789 ``` ### Rate Limiting Per-tier rate limits: - **Free**: 10 requests/minute - **Pro**: 100 requests/minute - **Enterprise**: 1000 requests/minute ### CORS Configuration Configured for major AI platforms: - Claude Desktop - Cursor IDE - Continue IDE - Codeium ## πŸ“Š Monitoring & Analytics ### Health Checks Monitor your deployment: ```bash # Check service health curl https://your-deployment.com/health # Check metrics curl https://your-deployment.com/metrics ``` ### Usage Analytics Built-in analytics track: - Request counts per user - Tool usage patterns - Error rates - Response times ## πŸŽ›οΈ Configuration ### Environment Variables #### Core Configuration ```bash NODE_ENV=production WORKSPACE_ROOT=/data ALLOWED_DIRS=/data:/tmp:/workspace ``` #### Auth & Payments (Optional) ```bash # Stripe Configuration STRIPE_SECRET_KEY=sk_live_... STRIPE_PUBLISHABLE_KEY=pk_live_... STRIPE_WEBHOOK_SECRET=whsec_... # Database Configuration SUPABASE_URL=https://... SUPABASE_ANON_KEY=eyJ... # Service URLs BASE_URL=https://your-domain.com MCP_PROXY_URL=http://mcp-proxy:8080 ``` ### Custom Configuration Modify `mcp-proxy-config.yaml` for custom settings (copy from `config/examples/mcp-proxy-config.yaml` to project root before building): ```yaml # Performance tuning performance: maxConnections: 50 requestTimeout: 300000 # Custom rate limits rateLimit: requests: 2000 window: 60000 ``` ## πŸ”§ Claude Desktop Integration Add your deployed service to Claude Desktop: ```json { "mcpServers": { "acf": { "url": "https://your-deployment.com/sse", "transport": "sse", "headers": { "Authorization": "Bearer YOUR_TOKEN_HERE" } } } } ``` ## πŸ› Troubleshooting ### Common Issues #### 1. Health Check Fails ```bash # Check logs docker logs acf-mcp-proxy # Verify configuration ./quick-deploy.sh test ``` #### 2. Authentication Errors ```bash # Verify token format echo "Bearer acf-demo-token-2024" | base64 # Check CORS settings curl -H "Origin: https://claude.ai" https://your-deployment.com/sse ``` #### 3. Rate Limiting ```bash # Check usage limits curl -H "Authorization: Bearer YOUR_TOKEN" \ https://your-deployment.com/metrics ``` ### Debug Mode Enable debug logging: ```bash # Local debugging DEBUG=true ./quick-deploy.sh local # Production debugging export DEBUG=true ``` ## 🚒 Production Deployment ### Security Checklist - [ ] Replace demo tokens with production tokens - [ ] Configure custom domain with SSL - [ ] Set up monitoring and alerting - [ ] Configure backup and disaster recovery - [ ] Review and tighten CORS settings - [ ] Set up log aggregation - [ ] Configure secrets management ### Performance Optimization 1. **Enable CDN**: Use CloudFlare for static assets 2. **Database Scaling**: Upgrade Supabase plan as needed 3. **Caching**: Enable Redis for token caching 4. **Load Balancing**: Scale horizontally with multiple instances ### Monitoring Setup 1. **Uptime Monitoring**: Set up Pingdom/StatusCake 2. **Error Tracking**: Configure Sentry 3. **Analytics**: Set up Google Analytics 4. **Business Metrics**: Track revenue and usage ## πŸ’‘ Next Steps ### Immediate (Day 1) 1. Deploy to your preferred platform 2. Test with Claude Desktop 3. Set up basic monitoring ### Short Term (Week 1) 1. Configure custom domain 2. Set up production authentication 3. Add monitoring and alerting ### Long Term (Month 1) 1. Optimize performance 2. Add custom features 3. Scale based on usage 4. Plan cloud-native migration ## πŸ†˜ Support - **Documentation**: Full docs in `/docs` - **Issues**: GitHub Issues for bug reports - **Enterprise**: Contact for enterprise support - **Community**: Discord server for discussions ## πŸ“œ License This deployment configuration is part of the Agentic Control Framework. See LICENSE for details. --- **Ready to deploy?** Choose your platform and run the quick-deploy script! πŸš€

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/FutureAtoms/agentic-control-framework'

If you have feedback or need assistance with the MCP directory API, please join our Discord server