Context Vault

OpenAI Codex Plugin for Claude Code

4 min read

OpenAI Codex Plugin for Claude Code — 경쟁사 에이전트 안에서 돌아가는 공식 플러그인

“We love an open ecosystem!”

— Dominik Kundel, OpenAI

Overview

OpenAI가 Claude Code용 Codex 플러그인을 공식 출시했습니다.

GitHub: https://github.com/openai/codex-plugin-cc

What is it?

Claude Code 안에서 Codex를 사용하는 방법:

  • 코드 리뷰 맡기기
  • 작업 통째로 넘기기
  • 백그라운드 작업 관리

Key Insight

경쟁사 에이전트 안에서 돌아가는 자기 에이전트.

Installation (3 Lines)

/plugin marketplace add openai/codex-plugin-cc
/plugin install codex@openai-codex
/reload-plugins

Then setup:

/codex:setup

If Codex is not installed:

npm install -g @openai/codex

Login if needed:

!codex login

Commands

/codex:review

Read-only Codex review on current work.

/codex:review
/codex:review --base main
/codex:review --background

/codex:adversarial-review

Steerable challenge review that questions the chosen implementation and design.

/codex:adversarial-review
/codex:adversarial-review --base main challenge whether this was the right caching and retry design
/codex:adversarial-review --background look for race conditions

Use when:

  • Reviewing design choices, tradeoffs, hidden assumptions
  • Pressure-testing risk areas (auth, data loss, rollback, race conditions)
  • Challenging the direction, not just code details

/codex:rescue

Delegate tasks to Codex through the codex:codex-rescue subagent.

/codex:rescue investigate why the tests started failing
/codex:rescue fix the failing test with the smallest safe patch
/codex:rescue --resume apply the top fix from the last run
/codex:rescue --model gpt-5.4-mini --effort medium investigate the flaky integration test
/codex:rescue --model spark fix the issue quickly
/codex:rescue --background investigate the regression

You can also just ask for a task to be delegated:

“Ask Codex to redesign the database connection to be more resilient.”

/codex:status

Shows running and recent Codex jobs for the current repository.

/codex:status
/codex:status task-abc123

/codex:result

Shows the final stored Codex output for a finished job.

/codex:result
/codex:result task-abc123

/codex:cancel

Cancels an active background Codex job.

/codex:cancel
/codex:cancel task-abc123

Features

Authentication

  • Uses your local Codex CLI authentication
  • Supports both ChatGPT account and API key
  • If already signed into Codex on this machine, it works immediately

Configuration

  • Uses same Codex install and config as direct usage
  • Project-level config in .codex/config.toml
  • User-level config in ~/.codex/config.toml

Example config:

model = "gpt-5.4-mini"
model_reasoning_effort = "xhigh"

Review Gate

Optional feature that runs a targeted Codex review before Claude completes a task.

/codex:setup --enable-review-gate
/codex:setup --disable-review-gate

Warning: The review gate can create long-running Claude/Codex loops and may drain usage limits quickly.

Philosophy: Open Ecosystem

“We love an open ecosystem!”

Key Principles

  1. Interoperability — Codex works with various dev environments and tools
  2. Flexibility — Developers use Codex within their preferred setups
  3. No Lock-in — No proprietary ecosystem restrictions

Supported Platforms

  • Claude Code
  • VSCode
  • Xcode
  • JetBrains
  • (And more via plugin ecosystem)

Technical Details

How it Works

  • Plugin wraps the Codex app server
  • Uses global codex binary in your environment
  • Applies same configuration as direct Codex usage
  • Delegates through local Codex CLI and Codex app server

What it Means

  • Same Codex install as direct usage
  • Same local authentication state
  • Same repository checkout and machine-local environment

Cost & Limits

  • Requires ChatGPT subscription (incl. Free) or OpenAI API key
  • Usage contributes to Codex usage limits
  • Pricing details

Use Cases

  1. Code Reviews — Get Codex-quality reviews without leaving Claude Code
  2. Bug Investigation — Delegate debugging to Codex in the background
  3. Refactoring — Let Codex handle complex code changes
  4. Adversarial Reviews — Challenge design decisions and assumptions
  5. Multi-Agent Workflows — Combine Claude Code’s local workflow with Codex’s cloud capabilities

Significance

This is remarkable because:

  1. OpenAI building for Claude Code — Direct competitor integration
  2. Official support — Not a third-party hack, but official OpenAI product
  3. Three-line install — Dead simple to add Codex to Claude Code workflow
  4. No lock-in — Users can choose the best tool for each task
  5. Ecosystem mindset — OpenAI embracing open ecosystem over walled garden

Source: 정석님 공유 (2026-03-30) + GitHub 리포지토리

그래프 뷰