Quick Start Guide¶
Get the MCP Gateway & Registry running in 5 minutes with this streamlined setup guide.
What You'll Accomplish¶
By the end of this guide, you'll have: - ✅ MCP Gateway & Registry running locally - ✅ Authentication configured with Amazon Cognito
- ✅ AI coding assistant (VS Code) connected to the gateway - ✅ Access to curated enterprise MCP tools
Prerequisites¶
- Amazon Cognito Setup: You'll need Cognito credentials (see minimal setup)
- Docker: Docker and Docker Compose installed
- Basic Command Line: Comfort with terminal/command prompt
Step 1: Clone and Configure¶
# Clone the repository
git clone https://github.com/agentic-community/mcp-gateway-registry.git
cd mcp-gateway-registry
# Copy and edit environment configuration
cp .env.example .env
Edit .env
with your values:
# Required - Replace with your actual values
COGNITO_USER_POOL_ID=us-east-1_XXXXXXXXX
COGNITO_CLIENT_ID=your_cognito_client_id
COGNITO_CLIENT_SECRET=your_cognito_client_secret
AWS_REGION=us-east-1
ADMIN_PASSWORD=your-secure-password
# Optional - Will be auto-generated if not provided
SECRET_KEY=optional-secret-key-for-sessions
Step 2: Generate Authentication¶
# Configure OAuth credentials for client access
cp credentials-provider/oauth/.env.example credentials-provider/oauth/.env
# Edit with minimal configuration
nano credentials-provider/oauth/.env
Add to credentials-provider/oauth/.env
:
# Ingress authentication (required for client access)
AWS_REGION=us-east-1
INGRESS_OAUTH_USER_POOL_ID=us-east-1_XXXXXXXXX
INGRESS_OAUTH_CLIENT_ID=your_cognito_client_id
INGRESS_OAUTH_CLIENT_SECRET=your_cognito_client_secret
Step 3: Install and Deploy¶
# Install Python environment
curl -LsSf https://astral.sh/uv/install.sh | sh
source $HOME/.local/bin/env
# Install Docker (Ubuntu/Debian)
sudo apt-get update
sudo apt-get install -y docker.io docker-compose
sudo usermod -a -G docker $USER
newgrp docker
# Deploy all services
./build_and_run.sh
⏱️ This takes about 2-3 minutes - Docker will build images and start services.
Step 4: Verify Installation¶
# Check all services are running
docker-compose ps
# You should see services like:
# - registry (port 7860)
# - auth-server (port 8888)
# - nginx (ports 80/443)
# - Various MCP servers (ports 8000-8003)
Access the web interface:
Login options: - Username: admin
(or your ADMIN_USER
value) - Password: Your ADMIN_PASSWORD
value
Step 5: Connect AI Coding Assistant¶
VS Code Setup (Recommended for first test)¶
# Copy generated VS Code configuration
cp .oauth-tokens/vscode-mcp.json ~/.vscode/settings.json
# If you have existing settings, merge instead:
cat .oauth-tokens/vscode-mcp.json >> ~/.vscode/settings.json
Test the Connection¶
- Open VS Code with MCP extension installed
- Open Command Palette (
Ctrl+Shift+P
orCmd+Shift+P
) - Run MCP command - you should see available MCP servers
- Try a tool - test with "current time" tool
Alternative: Roo Code Setup¶
Step 6: Test Everything Works¶
# Test gateway connectivity
cd tests
./mcp_cmds.sh ping
# Should return successful ping response
# Test specific tool
./mcp_cmds.sh call currenttime current_time_by_timezone '{"tz_name": "America/New_York"}'
Expected result: Current time in New York timezone
🎉 Success! What's Next?¶
You now have a fully functional MCP Gateway & Registry! Here are your next steps:
Immediate Next Steps¶
- 🔍 Explore the Web Interface - Browse available MCP servers and tools
- 🤖 Try AI Assistant Integration - Use tools through VS Code or your preferred AI assistant
- 🛠️ Add Your Own MCP Servers - Register custom tools for your team
Expand Your Setup¶
- 📚 Full Installation Guide - Production deployment options
- 🔐 Authentication Setup - Advanced identity provider configuration
- 🎯 AI Assistants Guide - Connect more development tools
Enterprise Features¶
- 👥 Fine-Grained Access Control - Team-based permissions
- 📊 Monitoring & Analytics - Usage tracking and health monitoring
- 🏢 Production Deployment - High availability and scaling
Amazon Cognito Minimal Setup¶
If you don't have Amazon Cognito configured yet, here's the minimal setup:
1. Create User Pool¶
# Using AWS CLI
aws cognito-idp create-user-pool \
--pool-name mcp-gateway-users \
--policies PasswordPolicy='{MinimumLength=8,RequireUppercase=false,RequireLowercase=false,RequireNumbers=false,RequireSymbols=false}' \
--region us-east-1
2. Create User Pool Client¶
# Create app client
aws cognito-idp create-user-pool-client \
--user-pool-id us-east-1_XXXXXXXXX \
--client-name mcp-gateway-client \
--generate-secret \
--explicit-auth-flows ADMIN_NO_SRP_AUTH CLIENT_CREDENTIALS \
--supported-identity-providers COGNITO \
--region us-east-1
3. Create Test User¶
# Create admin user
aws cognito-idp admin-create-user \
--user-pool-id us-east-1_XXXXXXXXX \
--username admin \
--temporary-password TempPass123! \
--message-action SUPPRESS \
--region us-east-1
# Set permanent password
aws cognito-idp admin-set-user-password \
--user-pool-id us-east-1_XXXXXXXXX \
--username admin \
--password YourSecurePassword123! \
--permanent \
--region us-east-1
For complete Cognito setup: See Amazon Cognito Setup Guide
Troubleshooting Quick Fixes¶
Services Won't Start¶
# Check Docker daemon
sudo systemctl status docker
sudo systemctl start docker
# Check port conflicts
sudo netstat -tlnp | grep -E ':(80|443|7860|8080)'
Authentication Errors¶
# Verify Cognito configuration
aws cognito-idp describe-user-pool --user-pool-id YOUR_POOL_ID
# Regenerate credentials
./credentials-provider/generate_creds.sh
Can't Access Web Interface¶
# Check if registry is running
curl http://localhost:7860/health
# Check logs
docker-compose logs registry
AI Assistant Not Connecting¶
# Verify configuration file exists
ls -la ~/.vscode/settings.json
# Test authentication manually
curl -H "Authorization: Bearer $(cat .oauth-tokens/ingress.json | jq -r .access_token)" \
http://localhost/mcpgw/sse
Getting Help¶
- 📖 Full Documentation - Comprehensive guides and references
- 🐛 GitHub Issues - Bug reports and feature requests
- 💬 GitHub Discussions - Community support and questions
- 📧 Troubleshooting Guide - Common issues and detailed solutions
🎯 Pro Tip: Once you have the basic setup working, explore the AI Coding Assistants Setup Guide to connect additional development tools like Cursor, Claude Code, and Cline for a complete enterprise AI development experience!