GitHub has officially introduced Agent HQ, a unified platform that integrates Anthropic’s Claude and OpenAI’s Codex directly into the developer workflow. Announced on February 4, 2026, this new capability allows developers to select specific AI agents for different coding tasks without leaving their primary workspace. The feature is currently available in public preview for GitHub Copilot Pro+ and Copilot Enterprise customers.
This update represents a significant shift in how developers interact with AI tools. Instead of relying solely on a single default model, users can now choose the agent best suited for a specific problem. Whether working on GitHub.com, the GitHub Mobile app, or Visual Studio Code, developers have immediate access to both Claude and Codex. This integration aims to reduce the friction caused by switching between different tools and interfaces, keeping all project context, history, and code reviews in one centralized location.
Transforming the Developer Experience
The core philosophy behind Agent HQ is flexibility. Developers often use different AI models for different stages of development, from brainstorming ideas to writing syntax. With this integration, a user can move from concept to implementation using different agents for each step. For example, a developer might use one agent to draft initial code and another to review it, all while maintaining the full context of the repository.
One of the most powerful features of this update is the ability to treat AI agents like human teammates. Users can assign GitHub issues directly to Claude, Codex, or Copilot using the standard “Assignees” dropdown menu. Once assigned, the selected agent automatically begins working on the task and submits a draft pull request for review. This workflow allows developers to iterate on code and provide feedback just as they would with a colleague, using comments to request changes or refinements.
Comparing Solutions with Multi-Agent Workflows
Agent HQ also introduces the ability to run agents in parallel, enabling developers to compare how different models approach the same problem. By assigning multiple agents to a single task, teams can see how Claude, Codex, and Copilot reason through tradeoffs and arrive at different solutions. This helps surface potential issues earlier in the development cycle.
GitHub suggests using this multi-agent capability for distinct types of review. For instance, one agent could be tasked with evaluating architectural guardrails to identify modularity issues. Simultaneously, a different agent could perform logical pressure testing to hunt for edge cases or scalability problems. A third agent might focus purely on pragmatic implementation, proposing backward-compatible changes to minimize risk. This method shifts the developer’s focus from writing syntax to evaluating high-level strategy and code health.
Access and Configuration
To access these new capabilities, users must have an active Copilot Pro+ or Copilot Enterprise subscription. No additional fees are required, as access to Claude and Codex is included in existing plans. During the public preview, each coding agent session consumes one premium request.
For individual users on Copilot Pro+, enabling these agents is done through the Copilot coding agent settings by toggling on specific models. Enterprise administrators have granular control over which agents are permitted across their organization. They can manage access policies in the “Enterprise AI Controls” section, ensuring that the use of external models aligns with company security standards.
For those working in Visual Studio Code, the feature requires version 1.109 or later. Developers can open the “Agent sessions” view to choose between local, cloud, or background tasks depending on their needs.
Future Expansions
While the current release focuses on Claude and Codex, GitHub has indicated plans to expand the roster of available agents. The company is actively working with partners such as Google, Cognition, and xAI to bring more specialized models into the GitHub ecosystem. These future integrations aim to provide even more specialized tools for developers, further cementing GitHub’s position as a central hub for AI-powered software development.
