How to Quickly Enable Codex in JetBrains IDEs and Boost Development

Why Codex in JetBrains IDEs?

IntelliJ IDEA 2024.3 introduced the built‑in AI Assistant, but developers found it slow and limited in complex scenarios. On January 22, OpenAI announced that the Codex extension now supports the entire JetBrains suite (Rider, IntelliJ IDEA, PyCharm, WebStorm, etc.), bringing stronger AI coding capabilities to familiar tools.

Getting Started with Codex

From JetBrains 2025.3, the Assistant can log in with an existing ChatGPT account. Open the Assistant sidebar, click the “Sign in Codex with ChatGPT” button, and follow the browser‑based authorization flow. If the Codex plugin is not installed, the IDE prompts you to install it. npm install -g @openai/codex In IntelliJ IDEA, search for the “Codex” plugin in the marketplace and install it.

Demo: Building a Project from Scratch

1. Automatic Project Retrieval

From GitHub download this project to the current directory: https://github.com/rymcu/mortise

Codex will automatically install Git if it is missing, achieving a “single command, fully automated” workflow.

2. Understanding an Unfamiliar Project

Introduce the project, describe each module’s purpose, outline its architecture, tech stack, and suggest how to get started quickly.

Codex analyzes the codebase, identifies the entry module, database, middleware, and dependencies, helping you grasp the project structure.

3. Automatic Environment Configuration

If the local environment lacks JDK, Maven, Docker, Redis, etc., Codex detects the gaps and installs the required tools. It can also switch JDK versions, e.g., installing JDK 21 when only JDK 17 is present.

4. Code Refactoring and Optimization

You can ask Codex to scan the whole project, list optimization opportunities by priority, or issue a direct command:

Without changing business logic, split complex logic and refactor the code.

5. Testing and Debugging

Codex infers boundary conditions from the business code, generates test cases covering normal and exceptional paths, runs them, and fixes failures until all tests pass. During debugging, it provides context‑aware root‑cause analysis and targeted solutions.

6. One Project per Day

Using Codex, the author completed several small projects in a single day, as illustrated by the following screenshots:

Beyond a Coding Assistant

Codex can also be combined with ChatGPT or Claude for broader tasks:

Documentation generation : automatically create README and API docs.

SQL and code generation : quickly produce common code snippets.

Front‑end development : use Codex with tools like Bolt to scaffold full applications (e.g., registration, payment, cloud storage).

Developers have reported completing an entire backend logic of a small program without writing a single line of code, achieving true “requirement‑driven development”.

Practical Advice

Network environment : When using Claude, a stable network is essential to avoid account bans.

Fundamental learning : AI does not replace programmers but can replace junior developers; solid fundamentals are still required to understand and verify AI‑generated code.