[FEATURE] Durable Agent

[hertaehyun]

Problem Statement

Durable agents are AI agents designed for production, offering resilience, state management, and cost-efficiency by persisting progress across failures (crashes, restarts) and long waits, unlike ephemeral agents.

Temporal (https://docs.temporal.io/) and DBOS (https://docs.dbos.dev/why-dbos) provide durable capability. Also, Microsoft Agent Framework provides kind of durable capability(https://learn.microsoft.com/en-us/agent-framework/user-guide/agents/agent-types/durable-agent/features?pivots=programming-language-python)

The Strands Agents provides significant benefits when using a durable agent, including crash recovery, automatic retries, idle agent suspension, simplified development, and enhanced observability.

Proposed Solution

AWS Lambda durable functions simplify building resilient, long-running, multi-step applications and AI workflows directly within the existing Lambda programming model. They allow functions to automatically checkpoint progress, suspend execution for up to one year without incurring compute charges, and recover from failures, all without requiring you to manage additional infrastructure.

By integrating with AWS Lambda, I hope Strands can provide durable capability.

Use Case

Use cases and industries requiring complex multi-agent collaboration patterns—such as Graph, Workflow, and Swarm—will benefit

Alternatives Solutions

No response

Additional Context

No response

[krokoko]

Thanks @hertaehyun , I was going to open the same issue :)

To add to this:

It seems different SDKs provide native integration with durable execution solutions:

• Pydantic AI supports temporal, DBOS, Prefect
• OpenAI SDK supports temporal
• Vercel has the Workflows SDK

It would be great to have some native integration within Strands.