OpenAI Just Shipped a Plugin for Claude Code — 25.8k Stars
OpenAI's official Codex plugin runs inside its competitor's harness and now tops GitHub trending. When your rival ships to your terminal, the terminal has become the platform.
NeuroX AI · July 6, 2026
OpenAI's newest developer tool doesn't run in ChatGPT. It runs inside Claude Code. The official codex-plugin-cc sits at the top of GitHub trending this week at 25.8k stars — an OpenAI-published plugin that lets you drive Codex from your competitor's harness without leaving the session.
The mechanics are one line: /plugin marketplace add openai/codex-plugin-cc. From there you get /codex:review for code review, background task delegation with /codex:rescue and /codex:transfer, and the command worth paying attention to — /codex:adversarial-review, which points a second frontier model at your diff with instructions to challenge the implementation.
Read the subtext. OpenAI looked at where engineering teams actually spend their day and concluded the answer was Anthropic's terminal agent. When a rival ships an official integration into your workflow layer, that layer has stopped being a tool and become a platform — and the model behind it becomes a swappable part.
The adversarial pattern is the production lesson. Teams that ship agent-written code reliably don't trust one model's opinion of its own work; they wire a second one to attack it before merge. That's cheap insurance a demo never includes — and exactly the kind of harness engineering that separates a prototype from something you can run unattended.