agentic-aiCommunity canon

Temporal

Durable execution engine for resilient, long-running agent workflows

View on GitHub

What it solves

The reliability problem in long-running agents: when an LLM call fails or a tool times out, Temporal retries from the last checkpoint automatically without you writing recovery logic.

Overview

Temporal is a workflow engine that makes code durable, if a process crashes, rate-limits, or throws an exception, Temporal replays the workflow from its last checkpoint automatically. For agentic systems, this means an agent loop that spans hours or days survives failures without you writing retry logic. Used in production at Stripe, Netflix, Datadog, and Coinbase. Official OpenAI Agents SDK integration launched September 2025.

Key facts

Language
Go
License
MIT
Maturity
Community canon
Maintainer
temporalio
Reviewed
2026-06-11

Where to start

Read the service README to understand Workflows, Activities, and the Event History, these three concepts explain why Temporal's durability guarantees are stronger than a simple retry loop.

service/README.md

Technologies

temporalgoworkflow

More in this domain