GitHub Adds Claude and Codex AI Coding Agents

GitHub has today integrated Anthropic's Claude and OpenAI's Codex AI coding agents directly within its platform. The new public preview brings Claude and Codex to GitHub, GitHub Mobile, and Visual Studio Code, available to users with a Copilot Pro Plus or Copilot Enterprise subscription.

This move is part of "Agent HQ," GitHub's initiative to natively embed AI agents into developers' daily workflows on GitHub. Developers can now choose between Copilot, Claude, Codex, or other custom agents when creating tasks. Each coding agent consumes a premium request, and developers can assign these agents to issues and pull requests.

Developers can also evaluate the performance of Copilot, Claude, and Codex, assessing the solutions generated by each AI coding agent. Mario Rodriguez, Chief Product Officer at GitHub, stated: "Context switching in software development creates friction." "With Codex, Claude, and Copilot in Agent HQ, you can use different agents for various steps—from idea to implementation—without switching tools or losing context."

GitHub is rapidly adopting competing AI models and agents to enhance its own Copilot product. Developers can already access models from Anthropic, Google, xAI, and OpenAI within GitHub Copilot, making the integration of rival AI coding agents a logical next step.

Access to Claude and Codex will soon be expanded to more GitHub Copilot subscription tiers. GitHub is also collaborating with Google, Cognition, and xAI to introduce more agents to GitHub, Visual Studio Code, and Copilot CLI.

GitHub's embrace of competing AI coding agents coincides with Microsoft increasingly trialing Anthropic's Claude Code tool. Microsoft's internal developers are being asked to compare Claude Code with GitHub Copilot to identify potential improvements for GitHub Copilot.