---
title: "How can I prevent duplicate tool actions when an OpenAI response is retried?"  
description: "How can I prevent duplicate tool actions when an OpenAI response is retried?"  
author: "Arnaud Simon"  
published: 2026-10-21  
updated: 2026-10-21  
canonical: https://www.mindstick.com/forum/162625/how-can-i-prevent-duplicate-tool-actions-when-an-openai-response-is-retried  
category: "Tool Calling"  
tags: ["OpenAI", "API Development", "Tool calling", "Backend Reliability"]  
reading_time: 1 minute  

---

# How can I prevent duplicate tool actions when an OpenAI response is retried?

I’m building an app where the model can ask our backend to create a support ticket or schedule a job. If a request times out after the model has returned a tool call, I’m not always sure whether the action ran or whether I should try again.

How are people handling **duplicate tool actions** in this situation? I’m especially interested in where you keep the execution record and whether an **idempotency key** is enough when the same tool call arrives more than once.


---

Original Source: https://www.mindstick.com/forum/162625/how-can-i-prevent-duplicate-tool-actions-when-an-openai-response-is-retried

Copyright © MindStick Software Pvt. Ltd. This Markdown version is provided for developers, AI systems, and offline reading.