GitHub is making Claude by Anthropic and OpenAI’s Codex AI coding agents directly available inside GitHub today. A new public preview adds Claude and Codex to GitHub, GitHub Mobile, and Visual Studio Code, for users with a Copilot Pro Plus or Copilot Enterprise subscription.

The move is part of Agent HQ, GitHub’s vision to make AI agents native to how developers use GitHub every day. Developers can now choose Copilot, Claude, Codex or other custom agents when they’re creating a task. Each coding agent will consume a premium request, and developers can assign agents to issues and pull requests.

Developers will also be able to judge how Copilot, Claude, and Codex perform, and weigh up how each AI coding agent has generated a solution. “Context switching equals friction in software development,” says Mario Rodriguez, chief product office at GitHub. “With Codex, Claude, and Copilot in Agent HQ, you can move from idea to implementation using different agents for different steps without switching tools or losing context.”

GitHub has been quick to embrace rival AI models and agents to improve its own Copilot offering. Developers are already able to access models from Anthropic, Google, xAI, and OpenAI in GitHub Copilot, so integrating in rival AI coding agents feels like a natural next step.

Access to Claude and Codex will expand to more GitHub Copilot subscription types soon, and GitHub is also working with Google, Cognition, and xAI to bring more agents into GitHub, Visual Studio Code, and the Copilot CLI.

GitHub’s embrace of rival AI coding agents comes as Microsoft is increasingly trialing Anthropic’s Claude Code tool. Developers inside Microsoft have been asked to compare Claude Code with GitHub Copilot, in an effort to improve GitHub Copilot.