Azure Durable Task

Build reliable, fault-tolerant workflows that survive any failure.

---

Table of Contents

• What is Durable Task?
• Why Durable Task Scheduler?
• Get Started in 5 Minutes
• Choose Your Framework
• Samples
• Observability
• API Reference
• AI-Assisted Development
• Contributing
• Community & Support

---

What is Durable Task?

Durable Task is Microsoft's technology for building workflows and orchestrations as ordinary code that automatically survives failures. Instead of managing complex retry logic, state machines, or message queues, you express your business logic as straightforward functions - Durable Task handles state persistence, automatic recovery, and distributed coordination for you.

Workflows can run for hours, days, or even months, reliably resuming from the last completed step after any crash, restart, or redeployment. Common use cases include distributed transactions, multi-agent AI orchestration, data processing pipelines, and infrastructure management.

Durable Task encompasses the Durable Task SDKs for self-hosted applications, Durable Functions for serverless hosting on Azure Functions, and the Durable Task Scheduler - a fully managed backend service purpose-built for durable workloads.

What is Durable Execution?

Durable execution is an industry-wide approach to making ordinary code fault-tolerant by automatically persisting its progress.

---

Why Durable Task Scheduler?

• 🏗️ Fully managed - no storage accounts to configure, no infrastructure to maintain
• ⚡ Purpose-built & fast - optimized compute+memory; push-model gRPC streaming (no polling)
• 📊 Built-in dashboard - monitor orchestrations, drill into history, pause/terminate/restart instances (Learn more)
• 🛡️ Fault isolation - runs as a separate Azure resource; failures don't cascade to your app
• 📈 Independent scaling - scheduler scales separately from your app; multiple apps can share one scheduler
• 🗂️ Multiple task hubs - isolate workloads by environment, team, or project (Learn more)
• 🐳 Emulator for local dev - Docker-based emulator with dashboard included, zero Azure dependency
• 🔐 Identity-based auth - Microsoft Entra ID / managed identity, no secrets in connection strings (Learn more)
• 🌍 Run anywhere - Azure Functions, Container Apps, AKS, App Service, VMs

---

⚡ Get Started in 5 Minutes

Step 0: Clone the repo

git clone --recurse-submodules https://github.com/Azure-Samples/Durable-Task-Scheduler.git
cd Durable-Task-Scheduler

Note: The --recurse-submodules flag is needed to pull sample code from linked repositories. If you already cloned without it, run: git submodule update --init --recursive

Step 1: Start the emulator

docker pull mcr.microsoft.com/dts/dts-emulator:latest
docker run -d -p 8080:8080 -p 8082:8082 mcr.microsoft.com/dts/dts-emulator:latest

Step 2: Pick your language and run a sample

Step 3: Open the dashboard

Navigate to http://localhost:8082 to view orchestration status, history, and more.

---

Choose Your Framework

📖 Choosing an orchestration framework →

---

Samples

Explore production-ready examples across languages and frameworks.

📂 Full Sample Catalog →

Featured Samples

🤖 AI Research Agent - Autonomous academic research agent that searches arXiv, analyzes papers, and synthesizes reports (Python)

✈️ AI Travel Planner - Multi-agent travel planning with specialized agents and human approval (Durable Functions, .NET)

🛒 Order Processor - End-to-end order workflow with inventory, payment, and notifications (Durable Functions, .NET)

🔄 Saga Pattern - Distributed transactions with compensating actions for failure recovery (Durable Functions, .NET)

🧩 Durable Extension for Microsoft Agent Framework - Make any Microsoft Agent Framework agent durable with persistent sessions, multi-agent orchestrations, and graph-based workflows (.NET, Python)

---

Observability

The Durable Task Scheduler provides a built-in dashboard for monitoring orchestration instances, inspecting execution history, and managing running workflows. The SDKs also support OpenTelemetry distributed tracing for end-to-end visibility across services. Durable Functions users can leverage distributed tracing V2 for enhanced diagnostics.

---

API Reference

Durable Task SDKs

• .NET
• Python
• Java
• JavaScript (coming soon)

Durable Functions

• .NET (isolated)
• Python
• Java
• JavaScript

---

AI-Assisted Development

This repository includes specialized skills for AI coding assistants (GitHub Copilot, Claude Code) to help you build durable workflows with best practices, code patterns, and contextual guidance.

Usage: Reference a skill file in your AI assistant (e.g., #file:.github/skills/durable-task-dotnet/SKILL.md in Copilot Chat) or ask it to read the skill before generating code. Skills are automatically detected by Claude Code when working on relevant files.

---

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines on how to get involved, submit issues, and open pull requests.

---

Community & Support

• 📖 Official Documentation
• 💬 GitHub Issues - bugs and feature requests
• 📧 Contact: nicholas.greenfield@microsoft.com, jiayma@microsoft.com

---

License

This project is licensed under the MIT License.

⭐ Star this repo if you find it useful!