Introducing Agent HQ: Any agent, any way you work

The present-day AI environment poses a familiar challenge: remarkable capabilities that are scattered across various tools and platforms. At GitHub, we’ve consistently aimed to tackle these systemic issues by enhancing accessibility through Git, establishing a structured code review process via pull requests, and streamlining deployments using Actions.

With a staggering 180 million developers, GitHub is expanding at an unprecedented rate, with a new developer joining every second. Even more impressive, around 80% of newcomers are embracing Copilot in their first week. AI has evolved from just a tool to a pivotal element of the development journey. It’s our duty to make sure this new collaborative era is not only robust and secure but also integrates seamlessly into the workflow you already rely on.

At GitHub Universe, we’re thrilled to introduce Agent HQ, which represents our vision for the next phase of our platform. Agents should not merely be an add-on; they should blend into the way you already work. That’s why we’re embedding agents directly into the GitHub workflow.

Agent HQ reimagines GitHub as an open ecosystem, bringing all agents together on a unified platform. In the coming months, coding agents from Anthropic, OpenAI, Google, Cognition, xAI, and many others will be accessible directly through GitHub as part of your paid Copilot subscription.

To bring this vision alive, we’re rolling out a suite of new features based on the fundamentals you trust. This starts with a mission control, a central hub to assign, manage, and monitor the tasks of multiple agents from anywhere. It extends to VS Code, where you’ll find new ways to plan and adjust agent behaviours. Furthermore, it’s reinforced by enterprise-level functionality: a fresh generation of agent-driven code reviews, a dedicated control plane to oversee AI access and agent functionality, and a metrics dashboard to analyse how AI impacts your work.

We’re firmly committed to investing in our platform and strengthening the basics you depend on every day. This new era of development is built on that foundational work, and we can’t wait to share more updates.

Let’s dive in.

GitHub is your Agent HQ: An open ecosystem for all agents

The future is about empowering you to coordinate a range of specialized agents to tackle complex tasks simultaneously, rather than juggling an assortment of disjointed tools or depending on a single agent. As forerunners in asynchronous collaboration, we see it as our obligation to ensure that these next-gen async tools simply work.

With Agent HQ, what remains unchanged is just as crucial as what’s new. You’re still using the basic tools you know—Git, pull requests, issues—and you can work with your preferred computation methods, whether that’s GitHub Actions or self-hosted runners. Accessing agents is straightforward through your existing paid Copilot subscription.

Additionally, we’re opening avenues to a fresh realm of capabilities. In the coming months, coding agents from Anthropic, OpenAI, Google, Cognition, and xAI will be available on GitHub as part of your paid GitHub Copilot subscription.

Are you eager for a head start? From this week, Copilot Pro+ users can begin using OpenAI Codex in VS Code Insiders, the first of our partner agents to go beyond its native environments and directly support your editor.

Mission control: Your command center, wherever you build

The strength of Agent HQ lies in mission control, a central hub designed to follow you wherever you work. It’s not confined to one location; it offers a consistent interface across GitHub, VS Code, mobile, and the CLI, enabling you to direct, monitor, and manage every AI-driven task. With mission control, selecting from a fleet of agents, assigning work in unison, and tracking their progress on any device is a breeze.

We’re also rolling out:

• New branch controls that offer detailed oversight on when to activate CI and other checks for agent-generated code.
• Identity features that allow you to specify which agent is performing the tasks, managing access and policies as you would for any other team member.
• One-click merge conflict resolution, improved file navigation, and enhanced code commenting features.
• Integrations for Slack and Linear, on top of our recent partnerships with Atlassian Jira, Microsoft Teams and Azure Boards, and Raycast.

Try mission control today.

New features in VS Code: Plan, customise, and connect

Mission control is integrated into VS Code as well, allowing a unified view of all your agents operating in VS Code, the Copilot CLI, or on GitHub.

Today’s fresh release in VS Code focuses on collaborating with agents throughout projects. We know that successful outcomes begin with a clear plan. It’s essential to have the right context upfront, and this context must persist as you progress. Copilot already tailors its functions to fit your team’s workflow by learning from your files and project culture. However, sometimes you require a more precise approach.

Introducing Plan Mode: this feature engages with Copilot by asking clarifying questions to help you develop a step-by-step strategy for your tasks. By providing context early on, you boost Copilot’s effectiveness and discover any gaps or unclear decisions before writing your code. Once you’ve finalised your plan, it goes to Copilot for implementation, whether locally in VS Code or with a cloud agent.

For even greater control, you can now create custom agents in VS Code using AGENTS.md files. These source-controlled documents allow you to set clear guidelines, like “favour this logger” or “utilise table-driven tests for all handlers.” This shapes Copilot’s behaviour without needing to re-prompt it each time.

Additionally, you can now leverage the new GitHub MCP Registry, accessible directly within VS Code. VS Code stands out as the only editor that fully supports the MCP specification. Discover, install, and activate MCP servers like Stripe, Figma, Sentry, and others, all with a single click. When specific expertise is needed for your task, you can create custom agents in GitHub Copilot with unique system prompts and tools to direct how you want Copilot to assist.

Boosted confidence and control for your team

Agent HQ not only empowers you but also builds your confidence. Ensuring code quality, understanding how AI influences your workflow, and maintaining oversight of AI interactions with your codebase and organisation are critical for your team’s prosperity, and we’re tackling these issues head-on.

When it comes to code quality, the key issue is that “LGTM” doesn’t always mean “the code is healthy.” A review might pass, yet still harm the codebase and lead to long-term technical debt. With GitHub Code Quality, available in public preview today, you gain organisation-wide visibility, governance, and reporting to systematically enhance code maintainability, reliability, and test coverage across all repositories. Implementing it broadens Copilot’s security checks to assess the maintainability and reliability impacts of modified code.

Additionally, we’ve introduced a code review phase within the Copilot coding agent’s workflow, ensuring that Copilot undergoes an initial review and addresses issues before you even see the code.

Your organisation needs to be aware of how Copilot is utilized. Today, we’re launching the public preview of the Copilot metrics dashboard, offering insights into Copilot’s impact and essential usage metrics across your entire organisation.

For enterprise administrators overseeing AI access—including AI agents and MCP—we’re focused on delivering consistent AI controls for teams through the control plane—the governance layer for your agents. Set security policies, maintain audit logs, and manage access all from one central location. Admins can also determine which agents are permitted, establish access to models, and gather metrics regarding Copilot usage in your organisation.

For developers, created by developers

We designed Agent HQ because we are developers, too. We understand how it feels when tools seem to hinder rather than assist. When “AI-powered” often means more context-switching, extra supervision, additional subscriptions, and more time clarifying your needs to gain the promised value.

That ends today.

Agent HQ isn’t just about AI hype; it’s about the tangible process of delivering code. It’s about instilling order and governance in this new era without losing choice. It’s about empowering you to build swiftly, with greater assurance, and on your own terms.

Welcome home. Let’s create together.

---