GitHub has officially expanded its AI capabilities with the launch of “Agent HQ,” a new platform feature that integrates Anthropic’s Claude and OpenAI’s Codex directly into the developer workflow. Announced in early February 2026, this update allows GitHub Copilot Pro+ and Enterprise users to select and run different AI coding agents without leaving their primary environment. By bringing these third-party models into GitHub and Visual Studio Code, the company aims to reduce context switching and provide developers with a broader range of tools for building software.
The introduction of GitHub Agent HQ marks a significant shift from a single-model approach to a multi-agent ecosystem. Developers can now utilize Claude and Codex alongside GitHub Copilot to tackle complex coding tasks, automate reviews, and accelerate implementation. These agents are currently in public preview and are accessible across GitHub.com, GitHub Mobile, and Visual Studio Code (version 1.109 and later).
The Power of Choice in Software Development
The core philosophy behind Agent HQ is the recognition that different AI models excel at different tasks. Rather than relying on a “one-size-fits-all” solution, developers can now choose the specific agent that best suits their immediate needs. For instance, a developer might employ Claude to evaluate architectural modularity and identify potential side effects, while simultaneously using Codex to propose a pragmatic, backward-compatible code change.
This flexibility allows for a more nuanced approach to problem-solving. Users can assign the same task to multiple agents to compare how they reason about tradeoffs and arrive at different solutions. This “logical pressure testing” helps surface edge cases and async pitfalls early in the development cycle, long before the code reaches production. By operating directly within the repository, these agents maintain the full context of the project, including history and previous reviews, eliminating the need to copy and paste code between external chat interfaces and the editor.
Streamlined Workflows in VS Code
For developers working in Visual Studio Code, the integration offers a unified “Agent Sessions” view. This interface serves as a command center where users can manage local, cloud, and background agents in one place. The update distinguishes between different types of agent interactions:
- Local Agents: Designed for fast, interactive help where the developer needs immediate feedback.
- Cloud Agents: Suited for autonomous tasks that run on remote infrastructure, allowing the developer to offload longer-running processes.
- Background Agents: Asynchronous tasks that run locally without blocking the main workflow.
A key advancement in this release is the ability to orchestrate subagents. Developers can fire off multiple context-isolated subagents in parallel to perform distinct tasks, such as researching documentation, scanning for security vulnerabilities, or checking authentication patterns. These subagents run independently, keeping the main session’s context clean and returning only the final result to the user.
Automating Issues and Pull Requests
Beyond the code editor, GitHub Agent HQ deeply integrates AI into project management and collaboration tools. Developers can now 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 extends to the review process itself. Agents can comment on pull requests, respond to feedback, and iterate on their own work until the task is complete. Developers can interact with these agents using mentions like @claude or @codex within the comment threads, treating the AI partners much like human teammates. This capability is designed to speed up the “idea to implementation” cycle by automating the initial drafts and routine code adjustments.
Enterprise Control and Security
With the introduction of multiple third-party agents, GitHub has implemented robust controls for enterprise environments. Administrators have the ability to define exactly which agents and models are permitted across their organization, ensuring that AI usage aligns with company security policies.
To maintain code quality, the platform includes automated checks that evaluate the maintainability and reliability of agent-generated code. Features like “GitHub Code Quality” and automated first-pass reviews allow the system to address initial problems before a human developer ever sees the code. Additionally, enterprise admins can track usage and impact through a dedicated metrics dashboard, providing traceability for all work generated by AI agents.
Expanding the Agent Ecosystem
This integration is built on open standards to foster a growing ecosystem of AI tools. Support for the Model Context Protocol (MCP) allows agents to return interactive UI components—such as dashboards and forms—directly within the chat interface, enabling richer collaboration. Furthermore, the general availability of “Agent Skills” allows extension authors to package and distribute specialized capabilities, such as testing strategies or API design patterns, which agents can then utilize.
While Claude and Codex are the first third-party additions, GitHub has announced plans to expand Agent HQ further. The company is actively working with partners like Google, Cognition, and xAI to bring more specialized agents into the platform in the near future.
