MEL Agent is an open-source platform for building, deploying, and managing AI-powered workflows. Think of it as n8n meets GPT – a visual workflow builder specifically designed for the AI era.

Unlike traditional automation platforms retrofitted for AI, MEL Agent is built from the ground up for LLM orchestration and AI agent workflows.

• Drag-and-drop interface inspired by n8n, Make.com, and Node-RED
• Real-time debugging with step-by-step execution traces
• Live data preview for every node in your workflow
• Built-in error handling and retry policies

• Multi-tenant architecture with row-level security (Opportunity for SaaS providers)
• SOC2 Type II compliance (Opportunity for enterprise adoption)
• GDPR compliant with automatic data lineage (Opportunity for EU markets)
• High availability for production workloads (Opportunity for critical systems)

• API-first design – everything in the UI is available via REST API
• Infrastructure as Code support (Opportunity for GitOps workflows)
• Extensive node library for common AI and automation tasks
• Custom node development with Go SDK

• 🧠 AI Nodes: LLM chat, embeddings (opportunity for image generation, speech-to-text)
• 🔗 Integration Nodes: Slack, Baserow, webhooks, HTTP requests (opportunity for Gmail, Notion, etc.)
• ⚡ Logic Nodes: If/else, transformations, delays, switches (opportunity for advanced loops)
• 📊 Data Nodes: Database queries, variable management (opportunity for file operations)
• 🔧 Utility Nodes: Logging, merging, splitting, workflow calls

• Envelope-based architecture for robust data propagation
• Automatic splitting and merging of array data (opportunity for advanced parallel processing)
• Global variables and context sharing across nodes
• Binary attachments support (opportunity for multimedia workflows)

• Live editing with WebSocket-based synchronization
• Version control with semantic versioning
• Execution history with full replay capability
• Team sharing and permission management (opportunity for enterprise collaboration)

MEL Agent follows modern cloud-native patterns with distributed workflow execution:

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│  React Frontend │    │   Go Backend     │    │   PostgreSQL    │
│                 │◄──►│                  │◄──►│                 │
│ • Visual Builder│    │ • REST API       │    │ • Multi-tenant  │
│ • Real-time UI  │    │ • WebSocket      │    │ • Encrypted     │
│ • Debug Tools   │    │ • Node Engine    │    │ • JSONB Storage │
└─────────────────┘    └──────────────────┘    └─────────────────┘
                                │
                                ▼
                    ┌─────────────────────────┐
                    │    Worker Architecture   │
                    ├─────────────────────────┤
                    │  📦 Local Workers       │
                    │  • Embedded in server   │
                    │  • Zero configuration   │
                    │  • Instant processing   │
                    ├─────────────────────────┤
                    │  🌐 Remote Workers      │
                    │  • Standalone processes │
                    │  • Horizontal scaling   │
                    │  • Geographic distribution │
                    │  • Load balancing       │
                    │  • Auto-registration    │
                    └─────────────────────────┘

MEL Agent supports both embedded and distributed execution:

• 🏠 Local Workers: Built-in workers that start with the API server for immediate processing
• 🌍 Remote Workers: Standalone worker processes that connect via secure API for scalable execution
• 🔄 Queue-based Coordination: Unified work queue system ensures reliable task distribution
• 💪 Fault Tolerance: Automatic retry, heartbeat monitoring, and worker health tracking

Technology Stack:

• Backend: Go with Chi router, PostgreSQL, WebSockets, Testcontainers
• Frontend: React, Vite, Tailwind CSS, ReactFlow
• Workers: Local embedded + distributed remote workers with queue-based coordination
• Infrastructure: Docker, Docker Compose, Kubernetes-ready

git clone https://github.com/cedricziel/mel-agent.git
cd mel-agent
docker compose up --build

🎉 That's it! Your MEL Agent instance is running:

• API: http://localhost:8080
• Web UI: http://localhost:5173
• Health Check: http://localhost:8080/health

Scale to multiple instances for production workloads:

# Start with load balancer and multiple instances
docker-compose -f docker-compose.scale.yml up -d --scale api=3 --scale worker=2

# Test scaling
./test-scaling.sh

🚀 Scaled deployment includes:

• Load Balancer: http://localhost:8080 (Nginx)
• 3 API Server Instances: Stateless, auto-scaling
• 2 Worker Instances: Distributed processing
• Health Monitoring: /health and /ready endpoints

Prerequisites: Go 1.21+, Node.js 18+, PostgreSQL 15+

# Backend (includes local workers)
export DATABASE_URL="postgres://postgres:postgres@localhost:5432/melagent?sslmode=disable"
go run ./cmd/server server

# Frontend
cd web
pnpm install && pnpm dev

# Optional: Remote Workers (for scaling)
export MEL_WORKER_TOKEN="your-worker-token"
go run ./cmd/server worker --token your-worker-token --server http://localhost:8080

• 🏗️ Architecture Overview
• 🔌 Connections & Integrations
• 🎨 Visual Builder Guide
• 📊 Data Flow & Envelopes

• 🧩 Building Custom Nodes
• 🔧 CLI Usage Guide
• 🚀 Deployment Guide
• 📈 Horizontal Scaling Guide

• 📧 Email Processing Agent (opportunity for real-world examples)
• 📱 Social Media Manager (opportunity for marketing automation)
• 📊 Data Pipeline Automation (opportunity for ETL workflows)

MEL Agent stands on the shoulders of giants, drawing inspiration from:

• n8n – Visual workflow automation and node-based architecture
• Make.com – Intuitive drag-and-drop interface and robust integrations
• Node-RED – Flow-based programming concepts and real-time debugging
• Zapier – Ease of use and extensive integration ecosystem

What makes MEL Agent different:

• ✨ AI-first design – Built specifically for LLM and AI agent workflows
• 🏗️ Modern architecture – Go backend, React frontend, envelope-based data flow
• 🔧 Developer-friendly – Comprehensive APIs, custom node SDK (opportunity for IaC support)
• 🏢 Open Source – Community-driven development, no vendor lock-in

We welcome contributions! MEL Agent is built by the community, for the community.

• 🐛 Report bugs via GitHub Issues
• 💡 Suggest features through GitHub Discussions
• 🔧 Submit PRs – see CONTRIBUTING.md for guidelines (opportunity to establish contribution guidelines)

# Backend
go test ./...              # Run tests (includes testcontainers)
go vet ./...              # Lint
go build ./cmd/server     # Build
go fmt ./...              # Format code

# Frontend
cd web
pnpm lint                 # Lint
pnpm build                # Build
pnpm test                 # Test (coming soon)
pnpm format               # Format code

# Server & Workers
./server --help                                    # Show all commands and options
./server server --help                             # Show server-specific help
./server server --port 8080                        # Start API server with embedded workers
./server api-server --help                         # Show api-server-specific help
./server api-server --port 8080                    # Start API server only (no embedded workers)
./server worker --help                              # Show worker-specific help
./server worker --token <token>                     # Start remote worker (basic)
./server worker --token <token> \                   # Start remote worker (advanced)
  --id worker-1 --concurrency 10 \
  --server https://api.example.com

# Configuration Examples
PORT=8080 ./server server                           # Use environment variables
./server server --port 9090                         # Override with flags
./server completion bash > /etc/bash_completion.d/mel-agent  # Install shell completion

MEL Agent is open source software licensed under the MIT License.

Special thanks to the open source community and the teams behind n8n, Make.com, Node-RED, and Zapier for pioneering visual workflow automation.

Ready to build your first AI agent? 🚀

Get started in 5 minutes → (opportunity for quick-start guide)