GitHub Copilot Adds Claude and Codex Agents

This multi-agent approach places GitHub in direct competition with AI-native IDEs like Cursor, which is built as a fork of VS Code and designed for a deeply integrated AI workflow. While Copilot operates as a plugin within various existing IDEs, Cursor offers a holistic environment with features like codebase-aware chat and advanced AI refactoring tools. This distinction is crucial for developers deciding between augmenting their current setup or adopting a fully AI-centric one. The integration of different AI models within a single platform reflects a growing trend of multi-tool AI workflows. Creatives and developers are increasingly chaining together specialized tools—like Midjourney for concept art, ChatGPT for copy, and now multiple coding agents—into cohesive pipelines. This interoperability allows practitioners to leverage the unique strengths of each model for different stages of a project, from ideation to implementation. This move also intensifies the philosophical debate around human-AI collaboration and authorship in creative and technical work. By allowing developers to switch between agents, the focus shifts from a single AI "assistant" to a team of specialized collaborators. This raises questions about creative agency: is the developer curating the output of different AIs, or is the AI a co-creator in the process? Anthropic's Claude models are known for their large context windows and strong performance in complex reasoning and coding tasks. OpenAI's Codex, the original model that powered the first version of GitHub Copilot, is recognized for its powerful code synthesis capabilities and has been trained on a massive dataset of public GitHub repositories. The ability to select different agents is managed through a unified platform within GitHub, ensuring shared context and memory across repository code, issues, and pull requests. Copilot Pro and Business users can enable Claude and Codex in their settings, and an admin must enable them at the enterprise and organization levels for business accounts. Each session with a partner agent, like Claude or Codex, consumes one premium request during the public preview. This integration is part of GitHub's broader "Agent HQ" vision, aiming to embed AI agents as a fundamental part of the daily development workflow. Developers can assign tasks to different agents, run them in parallel to compare approaches, and review their contributions in the form of draft pull requests and comments, much like interacting with human team members.