Codex CLI
Published Updated
What is Codex CLI?
Codex CLI is OpenAI’s terminal-based coding agent: it reads a local repository, edits files and runs commands using OpenAI’s GPT models. It is OpenAI’s counterpart to Claude Code.
An official terminal-based AI coding agent from OpenAI that reads local repositories, edits files, and executes commands autonomously.
Released as OpenAI's answer to Claude Code, the Codex CLI operates as an autonomous agent directly within the terminal. It leverages the latest GPT reasoning models to orchestrate complex refactors, run test suites, and manage local project files while integrating with local toolchains.
How it relates to Orbit
Orbit can register Codex as a coding harness and route a model tier to it. See Agent Sessions & Harnesses for details.
When you would use it
You reach for the Codex CLI when you are deeply embedded in the OpenAI ecosystem and want a powerful, terminal-first autonomous agent to execute software engineering tasks locally.
Common operations
- Orchestrating complex cross-file refactors locally.
- Automating testing and debugging directly from the command line.
Frequently asked questions
Do I need an OpenAI API key for Codex CLI?
You can sign in with a ChatGPT plan or use an API key. Orbit drives each agent through the login its CLI already has.
Can Orbit run Codex alongside Claude Code?
Yes. Both register as harnesses, and each complexity tier chooses its own agent, so one queue can send some prompts to Codex and others to Claude Code. See Complexity Tiers & Routing.
Can Orbit watch my Codex quota?
Yes. Orbit shows live provider quota for Claude, Antigravity and Codex, and can fall back to another model when one hits its limit. See Monitor Quotas & Cost.