Quickstart Guide

Get up and running with Duragraph in under 5 minutes! This guide will help you run your first AI workflow.

Prerequisites

Docker and Docker Compose
Python 3.8+ (for SDK usage)
Git

1. Start Duragraph Locally

Clone and start the Duragraph services:

git clone https://github.com/adwiteeymauriya/duragraph.git
cd duragraph
docker compose up -d

This starts:

API Server (port 8080) - REST API and SSE streaming
Temporal Server - Workflow orchestration engine
PostgreSQL - State and checkpoint storage
Workers - Python/Go execution environments

2. Install the Python SDK

pip install duragraph-sdk  # Coming in Q1 2024

3. Create Your First Workflow

Create a file called hello_workflow.py:

from duragraph import DuragraphClient, Workflow

# Initialize client
client = DuragraphClient(base_url="http://localhost:8080")

# Define a simple workflow
workflow = Workflow()
workflow.add_step("greet", {
    "type": "llm_call",
    "model": "gpt-3.5-turbo",
    "messages": [
        {"role": "user", "content": "Say hello to the world!"}
    ]
})

# Create and run the workflow
run = client.create_run(
    assistant_id="hello-assistant",
    thread_id="hello-thread",
    workflow=workflow
)

# Stream results
for event in client.stream_events(run.id):
    if event.type == "message":
        print(f"🤖 {event.data.content}")

4. Run the Workflow

python hello_workflow.py

You should see:

🤖 Hello, world! 👋 It's great to be here!

5. Monitor Your Workflow

Open the Duragraph Dashboard at http://localhost:3000 to:

View workflow execution logs
Monitor performance metrics
Debug failed runs
Explore workflow graphs

Next Steps

🎉 Congratulations! You've successfully run your first Duragraph workflow.

Ready to build more complex workflows? Check out:

Installation Guide - Production deployment options
Tutorials - Step-by-step workflow building
LangGraph Migration - Migrate from LangGraph Cloud
Python SDK Reference - Complete SDK documentation

Troubleshooting

Services won't start?

# Check service status
docker compose ps

# View logs
docker compose logs api

Python SDK not found? The Python SDK is currently in development (Q1 2024). For now, you can interact with the REST API directly:

curl -X POST http://localhost:8080/runs \
  -H "Content-Type: application/json" \
  -d '{"assistant_id": "test", "thread_id": "test"}'

Need help? Join our community or report an issue.

Quickstart Guide

On this page