OpenAI Ships Codex Plugin for Claude Code — Putting Its Agent Inside a Rival's Tool
In a move that blurs the lines between competing AI coding ecosystems, OpenAI has released codex-plugin-cc — a plugin that lets Claude Code users invoke Codex directly from within Anthropic's coding agent. Built by OpenAI engineer Dominik and published on GitHub, the plugin supports three workflows: a standard Codex code review, an adversarial review mode designed to surface edge cases and blind spots, and a task hand-off mode that passes work to Codex as a background second-opinion agent. The result is a cross-company integration that puts OpenAI's model inside a competitor's tool — deliberately.
The timing is notable. Anthropic recently reported that paid Claude subscriptions more than doubled in 2026, making Claude Code's developer community one of the largest and most active in the space. Rather than treating that momentum as a threat, OpenAI appears to be leaning into it — positioning Codex as a best-in-class review and delegation engine that works wherever developers already are. It's a bet on ubiquity over lock-in, and a signal that the AI coding-agent wars may be heading toward interoperability rather than walled gardens.
For developers already living in Claude Code, the plugin offers a straightforward way to get a second model's perspective without switching context. For the broader industry, it raises an interesting question: if AI coding tools are willing to integrate with their rivals, what does that mean for the moats each company is trying to build?