Skip to content
← Back to dictionary

Codex

OpenAI's AI coding agent for the terminal. Understands your repo, writes and runs code autonomously.

Lv.3 BuilderAI Coding

In plain English

Like Claude Code but from OpenAI. Another senior developer in your terminal. Different personality, same idea: describe what you want, it builds it.

Real-world example

You tell Codex 'write tests for all the API endpoints in this project.' It reads your routes, generates test files, and runs them to make sure they pass.

Where this fits in your project

OpenAI's terminal-based AI coding agent. Same category as Claude Code and Gemini CLI.

Idea & Planning
Prototyping
Frontend / UI
Backend / API
Database & Storage
Authentication
Deploy & Hosting
Polish & DX
Scale & Optimize

When to use this

When you want a terminal AI agent and prefer OpenAI's models.

The situation

You want a terminal AI agent and want to compare options or prefer GPT models.

Good for

  • + Multi-file changes
  • + Code generation
  • + Testing
  • + Working with GPT models

Not ideal for

  • - People uncomfortable with terminals
  • - When you need a visual editor

Getting started

Install with npm: `npm install -g @openai/codex`, then run `codex` in your project directory.

Alternatives

Claude Code (Anthropic's equivalent)Gemini CLI (Google's equivalent)aider (open-source)