OpenAI Ships Codex Plugin for Claude Code, Putting Its Agent Inside a Rival's Tool

OpenAI has released an official open-source plugin that allows developers to run its Codex coding agent directly inside Anthropic's Claude Code command-line interface. The move, announced on March 30, 2026, marks a rare instance of a major AI company building native tooling for a direct competitor's platform.

Key Highlights

OpenAI published codex-plugin-cc on GitHub, adding three slash commands to Claude Code
The plugin lets developers get code reviews, adversarial design challenges, and task delegation from Codex without leaving Claude Code
Installation takes three commands and works with a free ChatGPT account or OpenAI API key

How It Works

The plugin integrates through Claude Code's plugin marketplace and adds three core commands:

/codex:review scans uncommitted changes or branch diffs and returns a structured code review
/codex:adversarial-review goes further by challenging design decisions and testing assumptions, functioning like a second engineer questioning your approach
/codex:rescue delegates entire tasks to Codex, such as debugging failing tests, fixing regressions, or picking up from a previous session

All commands support background execution, with /codex:status and /codex:result available to check progress.

Why OpenAI Built This

The strategic logic is straightforward: Claude Code has become the dominant AI-powered coding tool, generating an estimated $2.5 billion in annualized revenue for Anthropic and accounting for roughly 4 percent of all public GitHub commits, around 135,000 per day. Rather than compete solely on distribution, OpenAI chose to meet developers where they already work.

"We love an open ecosystem and Codex should work with you wherever you are," said Dominik Kundel from OpenAI.

Meanwhile, Codex has grown to 1.6 million weekly active users, more than tripling after the February launch of GPT-5.3 Codex. Enterprise customers including Cisco, Nvidia, and Rakuten are already using the platform.

Security Concerns

On the same day as the plugin launch, security researchers at Phantom Labs disclosed a critical vulnerability allowing command injection through manipulated branch names, which could potentially expose GitHub tokens. OpenAI responded quickly with a patch that improved input validation and shell escaping.

Industry Reactions

The developer community response has been mixed. Some found the situation ironic, with one developer calling it "hilarious" that OpenAI is now building infrastructure for its biggest competitor. Others praised the practical utility of being able to cross-check different AI models within a single workflow.

As one observer put it: "The real winners in AI might not be the models themselves but whoever controls the distribution layer."

What This Means

The plugin signals a broader shift in the AI industry toward interoperability over exclusivity. As AI coding tools become essential developer infrastructure, companies are realizing that ecosystem reach may matter more than walled gardens. For developers, the immediate benefit is clear: access to both Claude and Codex capabilities without switching contexts.

Source: OpenAI codex-plugin-cc