GitHub has announced that it’s bringing Anthropic’s Claude and OpenAI’s Codex directly into Copilot, allowing customers to select third-party coding agents inside the core development experience.
Copilot users on Business and Pro plans can now choose between GitHub’s default assistant, Claude, and Codex when working across repositories, pull requests and issues. The agents operate within the same project context, meaning developers can assign tasks, review changes and iterate without switching tools.
The news comes a few weeks after the Microsoft subsidiary first revealed that Claude and Codex would be available in public preview for higher-tier Copilot subscribers, including Pro+ and Enterprise users. That earlier announcement positioned multi-agent support as a limited rollout, giving larger customers early access to external models running inside GitHub’s Agent HQ experience.
A choice of agents in Copilot
GitHub entered the AI coding space in 2021 with the launch of GitHub Copilot, developed in partnership with OpenAI. At the time, Copilot functioned as a single integrated assistant embedded directly into editors such as Visual Studio Code.
Since then, Copilot has expanded beyond inline code completion into a broader assistant capable of generating pull request summaries, suggesting fixes and responding to natural-language prompts about repository content.
The introduction of Claude and Codex inside Copilot marks a further shift: rather than relying on a single underlying system, GitHub is now offering developers a choice of agents within the same interface.
This means developers can compare outputs from different systems without leaving their repository. Tasks such as drafting code, reviewing changes or answering questions about project structure can be routed to whichever agent the user selects.
Agents can also be assigned directly to issues and pull requests, where they generate draft, reviewable outputs inside the existing workflow.
Competitive pressure in AI coding
The decision reflects a crowded and fast-moving AI coding landscape. Developers increasingly test multiple assistants across different tools, and model providers are competing for mindshare within established platforms.
A similar shift is underway elsewhere. Apple, for instance, recently added direct support for Claude and Codex inside Xcode, allowing the agents to operate within its flagship development environment. That move, like GitHub’s, embeds third-party systems deeper into established tools, giving developers access to multiple AI agents without leaving their primary IDE.
There is also a strategic dimension for Microsoft. While Copilot has deep ties to OpenAI, integrating a model from Anthropic signals a willingness to accommodate rival systems where customer demand exists.
For Business and Pro subscribers, the immediate effect is fairly obvious: more choice inside the same development environment. For GitHub, the broader objective is to ensure that as AI coding agents proliferate, they do so within its platform rather than outside it.
How to enable and use Claude and Codex
For Copilot Pro users, Claude and Codex can be enabled directly from the Copilot coding agent settings. Developers select which repositories agents can access, then toggle Claude, Codex or both on.
For Copilot Business customers, access must first be enabled by an administrator at the enterprise level under Enterprise AI Controls → Agents, and then again at the organization level under Settings → Copilot → Coding agent.
Once enabled, agents can be launched from the Agents tab inside a repository on GitHub’s web interface or mobile app. Users select an agent from a dropdown, enter a request and submit it. Sessions run asynchronously, with progress logs visible during execution.
