Codex in JetBrains IDEs

OpenAI’s Codex is now integrated directly into JetBrains IDEs, allowing developers to use AI-powered coding assistance—such as debugging, implementing features, and localization—seamlessly within their preferred JetBrains environment using a ChatGPT subscription, API key, or JetBrains AI subscription. The integration is flexible, non-intrusive, and enhances productivity by enabling Codex to analyze project context, suggest code changes, and work in parallel with developers without disrupting their workflow.

OpenAI has announced that Codex, their AI coding assistant, is now integrated directly into JetBrains IDEs, one of the most requested features from developers. This integration allows users to collaborate with Codex inside their preferred JetBrains environment, such as IntelliJ, PyCharm, WebStorm, or Rider. Codex can be accessed using a ChatGPT subscription, an API key, or a JetBrains AI subscription, making it flexible for different user preferences.

During the demonstration, Gleb from JetBrains showcased how Codex operates within the IDE on a real-world, multi-platform Kotlin project. He highlighted that Codex can analyze the project structure, understand its context, and assist with tasks such as debugging and implementing new features. The integration is seamless, and users do not need a JetBrains AI subscription specifically; logging in with a ChatGPT account or API key is sufficient.

Gleb demonstrated Codex’s ability to help resolve build errors by simply pasting error messages into the Codex chat. Codex then analyzed the relevant files, provided reasoning for its actions, and made the necessary code changes. After Codex’s intervention, the project built successfully, illustrating how the AI can autonomously diagnose and fix complex issues without requiring detailed guidance from the user.

The video also explored Codex’s capabilities in handling more complex tasks, such as adding Spanish localization to the app. Codex researched the codebase, made the required changes, and verified its work by running build tools and checking the results in the iOS simulator. The integration offers different access modes, allowing users to control how much autonomy Codex has, from read-only to full access, ensuring safety and flexibility.

Throughout the process, Codex worked in parallel with the developer, allowing them to continue their tasks without interruption. The changes made by Codex could be reviewed and committed using standard IDE tools. The presenters emphasized that Codex’s integration feels native and non-intrusive, enhancing productivity without altering established workflows. To get started, users can visit openai.com/codex or enable Codex in the AI chat sidebar of their JetBrains IDE.