agent-framework-durabletask

What it is and what it does

This package wraps Microsoft Agent Framework agents to run them inside the Durable Task orchestration framework. Durable Task is a distributed orchestration system that automatically handles state persistence, replays conversation history on recovery, and manages failures transparently. You create a TaskHubGrpcWorker connected to a Durable Task server, wrap it in DurableAIAgentWorker, and register your agents with it; the framework then handles the rest.

The package is in beta and depends on agent-framework-core, durabletask, durabletask-azuremanaged, and python-dateutil. It targets Python 3.10 and later and is actively maintained by Microsoft. Use it when you need agent conversations to survive crashes, scale across multiple processes, or replay interactions for debugging.

Use it for:

• Host long-running conversational agents that must survive server restarts or crashes without losing state.
• Build multi-step agent workflows that orchestrate across multiple services and need automatic retry on failure.
• Replay agent conversations for debugging or auditing by replaying the same inputs through the durable history.
• Scale agent workloads across multiple worker processes using Durable Task's distributed orchestration.
• Integrate AI agents into existing Azure-managed Durable Task deployments for enterprise reliability.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Integrates Microsoft Agent Framework agents with the Durable Task framework to enable state persistence, conversation history replay, and automatic failure recovery.

Yes, if you are building agents with Microsoft Agent Framework and need durability, state persistence, and failure recovery. The package is actively maintained, has low install friction, carries no known vulnerabilities, and integrates tightly with the agent framework ecosystem. It is in beta, so expect potential API changes; verify compatibility with your target agent-framework-core version before committing to production.

Install

agent-framework-durabletask on PyPI

pip install agent-framework-durabletask

uv add agent-framework-durabletask

poetry add agent-framework-durabletask

Installing agent-framework-durabletask

Before you install

Low install friction with a pure-Python wheel. The package is actively maintained (last commit 2026-08-14) and in beta status. Four runtime dependencies are all from the Microsoft agent ecosystem, suggesting tight integration and coordinated maintenance.

License in practice

Licensed under MIT (permissive), so you can use, modify, and distribute the package freely with minimal restrictions.

Quickstart

pip install agent-framework-durabletask --pre

from agent_framework_durabletask import DurableAIAgentWorker
from durabletask.worker import TaskHubGrpcWorker

worker = TaskHubGrpcWorker(host_address="localhost:4001")
agent_worker = DurableAIAgentWorker(worker)

Requires Python 3.10 or later. A running Durable Task gRPC server (e.g., on localhost:4001) is needed for the worker to connect.

Verify before relying

• Whether the package works with agent-framework-core versions other than the one it was released with.
• Performance characteristics and scalability limits for managing multiple concurrent agents.
• Exact retry and failure-recovery semantics compared to other orchestration frameworks.

Package facts

Evidence: agent_framework_durabletask-1.0.0b260730-py3-none-any.whl

Tags

More Application Frameworks packages

Further reading