Agent HQ brings third-party coding agents into GitHub and VS Code

Copilot Pro+ and Copilot Enterprise users can now run multiple coding agents directly inside GitHub, GitHub Mobile, and Visual Studio Code, with Copilot CLI support planned. Available in public preview today are Claude by Anthropic and OpenAI Codex, alongside GitHub Copilot itself. The goal is to cut context-switching by keeping agent activity, history, and review attached to the repository rather than scattered across separate tools.

We're bringing Claude into GitHub to meet developers where they are. With Agent HQ, Claude can commit code and comment on pull requests, enabling teams to iterate and ship faster and with more confidence. Our goal is to give developers the reasoning power they need, right where they need it.

Katelyn Lesse, Head of Platform, Anthropic

Comparing agent approaches on the same task

Agent HQ also makes it practical to see how different models reason about the same problem. You can assign multiple agents to a single task and compare how Copilot, Claude, and Codex weigh tradeoffs and arrive at different solutions. That opens the door to layered review strategies:

  • Architectural guardrails: Have an agent evaluate modularity and coupling, flagging changes that might introduce unintended side effects.
  • Logical pressure testing: Use another agent to hunt for edge cases, async pitfalls, or scale assumptions that could fail in production.
  • Pragmatic implementation: Let a separate agent propose the smallest backward-compatible change to minimize the blast radius of a refactor.

Our collaboration with GitHub has always pushed the frontier of how developers build software. The first Codex model helped power Copilot and inspired a new generation of AI-assisted coding. We share GitHub's vision of meeting developers wherever they work, and we're excited to bring Codex to GitHub and VS Code. Codex helps engineers work faster and with greater confidence—and with this integration, millions more developers can now use it directly in their primary workspace, extending the power of Codex everywhere code gets written.

Alexander Embiricos, OpenAI

Why agents belong inside the repo

Running agents where code already lives means changes and discussion stay connected to the pull request or issue, instead of forcing developers to copy context between a stateless chat window and their editor. Agent-generated work arrives as draft pull requests and comments, reviewed the same way a teammate's contribution would be—no new dashboards or separate AI-centric processes to learn.

Org-wide control and observability

The feature set extends beyond the individual developer. Agent HQ offers systematic control over how AI touches the codebase:

  • Agent controls: Enterprise admins can manage access and security policies in one place, defining which agents and models are allowed across the organization.
  • Code quality checks: GitHub Code Quality, also in public preview, extends Copilot's security checks to assess maintainability and reliability impact of changed code.
  • Automated first-pass review: A code review step is integrated into Copilot's workflow, letting it catch initial problems before a developer ever looks at the code.
  • Impact metrics: The Copilot metrics dashboard (in public preview) tracks usage and impact org-wide, offering traceability for agent-generated work.
  • Security and auditability: Audit logging and enterprise-grade access management keep agent activity aligned with existing security posture.

What's next

Access to Claude and Codex is expected to expand to more Copilot subscription tiers. GitHub is also working with partners including Google, Cognition, and xAI to bring additional specialized agents into GitHub, VS Code, and Copilot CLI workflows.