DuraGraph

An open, extensible orchestration layer for AI and workflow automation

DuraGraph provides a LangGraph Cloud-compatible API built with Event Sourcing and CQRS patterns for reliable, observable, and maintainable AI pipelines that can be self-hosted in enterprise environments.

🎯 Mission

Enable reliable, observable, and maintainable AI pipelines that feel natural for developers—bringing the power of LangGraph Cloud to self-hosted and enterprise environments with:

• API Compatibility: Drop-in replacement for LangGraph Cloud APIs
• Enterprise Ready: Self-hosted, compliant, secure
• Fault Tolerant: Event sourcing with reliable event delivery via outbox pattern
• Observable: Rich monitoring and workflow introspection

🚀 Quick Start

Get started with DuraGraph in minutes:

📖 View Documentation | 🎓 Quick Start Guide

One-Click Deploy

Deploy DuraGraph to your preferred cloud platform:

Local Development

# Clone the repository
git clone https://github.com/Duragraph/duragraph.git
cd duragraph

# Start all services with Docker Compose
docker-compose up -d

# Or use Task
brew install go-task/tap/go-task
task up

Visit http://localhost:8080 for the API and http://localhost:5173 for the dashboard.

👉 Full Setup Guide

🏗️ Architecture

flowchart LR
  client["Client SDKs / LangGraph Cloud clients"]
  api[API Server - Go/Echo]
  eventstore[(Event Store - PostgreSQL)]
  nats[NATS JetStream]
  engine[Graph Execution Engine]
  dashboard[Svelte Dashboard]

  client --> api
  api --> eventstore
  api --> engine
  eventstore --> nats
  nats --> dashboard
  engine --> eventstore

Loading

🔧 Architecture Details

⚡ Key Features

• 🔄 LangGraph Cloud API Compatible - Drop-in replacement for existing LangGraph Cloud integrations
• 🏢 Self-Hosted - Full control over your data and infrastructure
• ⚡ Event Sourcing & CQRS - Reliable, auditable workflow execution with event-driven architecture
• 🔍 Observable - Rich monitoring, tracing, and debugging tools with Prometheus metrics
• 🧩 Extensible - Custom graph execution engine with support for LLM nodes and tool execution
• 📊 Visual Dashboard - Real-time workflow visualization with Server-Sent Events
• 🐳 Docker Ready - Easy deployment with Docker Compose or Kubernetes

📚 Documentation

• Getting Started - Installation and basic usage
• API Reference - Complete API documentation
• Architecture - System design and components
• Development Guide - Contributing and development
• Deployment - Production deployment guides
• Operations - Monitoring and maintenance

🔧 Basic Usage

from langgraph_sdk import get_client

# Point to your DuraGraph instance
client = get_client(url="http://localhost:8080")

# Use exactly like LangGraph Cloud
assistant = await client.assistants.create(...)
thread = await client.threads.create()
run = await client.runs.create(
    thread_id=thread["id"],
    assistant_id=assistant["id"]
)

📖 Full API Documentation

🗂️ Project Structure

duragraph/
├── cmd/server/          # API server (Go)
├── internal/
│   ├── domain/          # Domain models (aggregates, entities, events)
│   ├── application/     # Use cases (commands, queries, services)
│   ├── infrastructure/  # External concerns (HTTP, persistence, messaging)
│   └── pkg/             # Shared utilities (errors, eventbus, uuid)
├── dashboard/           # Svelte visualization dashboard
├── website/             # Landing page (Vite/React)
├── docs/                # Documentation (Fumadocs/Next.js)
├── deploy/              # Docker, SQL migrations
└── Taskfile.yml         # Development task runner

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-feature
3. Make your changes and add tests
4. Run tests: task test
5. Open a Pull Request

🛠️ Development Guide

📄 License

Licensed under the Apache License 2.0.

🙋 Support

• Documentation: duragraph.dev/docs
• Issues: GitHub Issues
• Discussions: GitHub Discussions

🗺️ Roadmap

• LangGraph Cloud-compatible API
• Event sourcing with CQRS pattern
• Custom graph execution engine
• Outbox pattern for reliable event delivery
• PostgreSQL event store with NATS JetStream messaging
• Fumadocs documentation site
• Svelte dashboard for visualization
• Server-Sent Events streaming
• Enhanced LLM provider support (additional models)
• Advanced workflow patterns (parallel execution, subgraphs)
• Production Helm charts
• Multi-tenant support
• Workflow versioning and migration tools

📋 Full Roadmap

---

DuraGraph - Bringing enterprise-grade AI workflow orchestration to everyone.

Get Started · Documentation · Community