Revolutionizing Development: GitHub Copilot's Agent Mode & New Features in VS Code

GitHub Copilot is stepping up its game in Visual Studio Code, and the news is as exciting as it is transformative. In celebration of Microsoft’s 50th anniversary—a true milestone for a company that’s been at the heart of developer culture for decades—the latest wave of innovations is rolling out to all VS Code users, featuring Agent Mode with Model Context Protocol (MCP) support. These enhancements, along with the introduction of a new Copilot Pro+ plan and advanced premium model requests, are set to supercharge developer productivity and make coding more intuitive than ever before.

Agent Mode in Visual Studio Code: A New Chapter​

Agent Mode is not just another feature—it represents a paradigm shift in how developers interact with their coding assistants. Traditionally, Copilot has been a reactive tool, answering questions and offering inline code completions. Now, with Agent Mode, it becomes proactive. Instead of merely suggesting code, it can perform a series of actions across multiple files to bring your ideas to life. Imagine having an assistant that not only helps you write code but also troubleshoots, automates refactoring, and even suggests terminal commands when errors occur.
Some of the standout capabilities of Agent Mode include:

• Automatically identifying and handling subtasks to achieve your primary coding goal.
• Generating multi-file edits through a conversation-style prompt.
• Analyzing runtime errors with self-healing capabilities that adapt as you code.
• Suggesting terminal commands or tool calls to streamline your workflow.

Developers who have already been testing Agent Mode in the VS Code Insiders have reported a range of applications—from automatically fixing code generation errors to even “yeeting commits” (yes, you read that right). This innovative approach is underpinned by a multi-model architecture that brings in some of the most advanced large language models available today.

Key Takeaways:​

• Agent Mode transforms the way you interact with Copilot by automating multi-file edits.
• It’s designed to understand and execute a series of tasks, bridging the gap between question-answering and actionable coding.
• Self-healing capabilities allow it to adapt to errors in real-time, offering a more streamlined development process.

Multiplying Your Coding Arsenal: Multi-Model Support & Premium Requests​

If you’ve ever wished your code assistant could do it all, the new multi-model support and premium request system are here to answer that call. GitHub Copilot now integrates top-tier models from Anthropic, Google, and OpenAI, ensuring you have access to the best tool for every task. The selection of models now includes Anthropic’s Claude 3.5 and 3.7 Sonnet (as well as Sonnet Thinking), Google’s Gemini 2.0 Flash, and OpenAI’s leading model for code completions and context-driven tasks.
This multi-model flexibility means you can choose the model that best suits your project’s needs without compromising on quality or performance. And with premium requests, Copilot users can now leverage these models in ways that go beyond the unlimited calls available for the base model:

• Copilot Pro users will receive 300 monthly premium requests beginning May 5, 2025.
• Copilot Business and Enterprise users get 300 and 1000 monthly premium requests, respectively, with a staggered rollout between May 12 and May 19, 2025.
• The new Copilot Pro+ plan, designed for individual developers, delivers up to 1500 monthly premium requests and provides access to cutting-edge models like GPT-4.5 for just $39 per month.
• For those who need more flexibility, a pay-as-you-go option is available at approximately $0.04 per additional premium request.

This tiered system is designed to ensure that whether you’re a hobbyist, a freelance developer, or part of a large enterprise, you can harness the power of advanced AI without worrying about hitting a cap.

Highlights:​

• Multiple top-tier models ensure that you have the right tool for every aspect of development.
• Premium requests add an additional layer of capability, allowing for more complex operations.
• New pricing tiers and usage options mean that advanced AI support is accessible to both individual developers and larger teams.

Unlocking New Capabilities with the Model Context Protocol (MCP)​

At the heart of these enhancements is the Model Context Protocol (MCP), a new way for your AI assistant to interact with your development environment. Think of MCP as a “USB port for intelligence”—a seamless interface that connects Agent Mode to the tools and services you rely on. With MCP, the AI isn’t just passively reading your code; it’s actively tapping into databases, querying web services, and integrating with a robust GitHub ecosystem.
The MCP support in Agent Mode enables a host of capabilities:

• It allows your Copilot to fetch contextual information from various sources—a boon for tasks that require up-to-date data, such as updating your GitHub profile or managing repository issues.
• By using the GitHub local MCP server (available as an open-source project), developers can integrate powerful GitHub functionalities directly into their coding tasks. This includes searching across repositories, managing issues, and even creating pull requests without leaving VS Code.
• The iterative tool calling facilitated by MCP ensures that the AI can continuously refine its actions based on real-time feedback, making it not only reactive but intelligently proactive.

For example, if you prompt Agent Mode to “Update my GitHub profile to include the title of the PR that was assigned to me yesterday,” the tool can leverage MCP to call various services, gather the necessary context and execute the update across your profile seamlessly. This level of integration demonstrates just how far AI-driven workflows have come.

MCP in a Nutshell:​

• MCP acts as a bridge between your AI assistant and your wider development ecosystem.
• It facilitates seamless interactions with external tools, enhancing the contextual understanding of Agent Mode.
• The open-source GitHub local MCP server puts powerful repository management features at your fingertips.

Behind the Agent Mode: How It Works and What It Means for Developers​

Understanding Agent Mode is like peering under the hood of a high-performance engine. At its core, Agent Mode leverages chain-of-thought reasoning to break down your commands into actionable steps. Instead of handling a single line of code or a simple question, it dissects your prompt into multiple sub-tasks, each executed across different files and contexts. This approach has already proven effective in early testing environments, with a pass rate of 56.0% on SWE-bench when verified with Claude 3.7 Sonnet.
The operational workflow is both intuitive and powerful:

• You provide a prompt or a high-level description of what you want to achieve.
• Agent Mode identifies the necessary subtasks required to meet that goal.
• The AI then iteratively calls on external tools and services (thanks to MCP) to execute each task.
• It monitors the outputs, refines its actions, and addresses any runtime errors with self-healing logic.

This process is akin to having a team of experts working in tandem—each one specializing in a different aspect of the development cycle—while still being orchestrated by a single, intelligent interface.

How Agent Mode Enhances Productivity:​

• Streamlined multi-step coding workflows reduce the time spent on mundane tasks.
• Self-healing capabilities minimize downtime when errors occur.
• Automated terminal command suggestions and tool integrations speed up routine operations.
• Overall, it aims to let you focus on creativity while handling the grunt work behind the scenes.

Agent Mode is set to transform coding from a linear, file-by-file process into a dynamic, context-rich experience that feels almost conversational.

Boosting Developer Productivity with Code Review and Next Edit Suggestions​

Alongside Agent Mode, GitHub is rolling out two more standout features that promise to redefine developer workflows: the Copilot code review agent and next edit suggestions.

Copilot Code Review Agent​

Previously in preview, the Copilot code review agent has now graduated to general availability, having already assisted over 1 million developers on GitHub. This tool takes a proactive role in your development process by:

• Analyzing code for potential issues before they escalate.
• Suggesting improvements and refinements based on best practices.
• Acting as an automated peer reviewer, providing insights that can help maintain code quality across large projects.

The integration of this tool within VS Code means that code reviews become less about manual back-and-forth and more about instant, on-the-fly feedback that can elevate the overall standard of your development work.

Next Edit Suggestions​

Next edit suggestions are yet another feather in the cap of GitHub Copilot’s new functionality. Instead of waiting for a pull request or afterthought during a lengthy coding session, now you can simply hit Tab—repeatedly if needed—and watch as Copilot offers context-aware code completions that evolve with your project. This proactive approach not only accelerates the coding process but also helps reduce the friction of switching between writing and editing modes.

Productivity Enhancements Summarized:​

• Code review agent automates the feedback process, ensuring high-quality code with minimal manual oversight.
• Next edit suggestions streamline the editing workflow, allowing for more natural and fluid coding sessions.
• Both features, powered by advanced AI and integrated directly within VS Code, ensure you spend more time building and less time debugging.

Getting Started with the New Features in VS Code​

Integrating these improvements into your development process is straightforward:

• Upgrade to the latest version of Visual Studio Code.
• Navigate to your settings and enable Agent Mode to begin harnessing its full power.
• Explore the configuration options to link MCP tools and set up your local GitHub MCP server if needed.
• For those interested in the premium model requests, review your subscription options, be it Copilot Pro, Business, or Enterprise, and consider upgrading to the new Pro+ tier for unrivaled access to cutting-edge AI models.

These steps ensure that you not only get the latest functionalities but also tailor your coding environment to your specific workflow. With a simple upgrade and some configuration tweaks, you’re ready to dive into a more intelligent, intuitive coding experience.

Quick Setup Checklist:​

• Update VS Code to the latest version.
• Enable Agent Mode through the settings.
• Configure MCP integrations for enhanced tool context.
• Review your Copilot subscription options and consider premium request add-ons if needed.

Looking Ahead: A Future Fueled by Developer Empowerment​

Microsoft’s half-century mark is more than just a celebration—it’s a reflection of the company’s enduring commitment to the developer community. The rollout of Agent Mode, MCP support, premium model requests, the Copilot code review agent, and next edit suggestions all signal a future where coding is increasingly collaborative, dynamic, and context-aware.
The innovative multi-model support underscores Microsoft and GitHub’s vision to cater to a diverse array of development needs. By giving developers the freedom to leverage multiple advanced models, the platform intuitively adapts to various coding scenarios. Whether you are debugging complex issues, managing extensive code reviews, or simply writing new code, these tools are built to streamline your workflow and maximize efficiency.
Moreover, the integration of MCP ensures that your coding assistant isn’t working in isolation. It has full access to your development ecosystem—integrating databases, telemetry, and repository management—ensuring that every command or suggestion is contextually relevant and actionable. As chain-of-thought reasoning models continue to mature, expect even more sophisticated interactions, where the assistant not only anticipates your needs but also offers suggestions you might never have thought of yourself.

The Broader Impact:​

• GitHub Copilot’s advancements are emblematic of a broader shift in developer tools, where AI is no longer passive but an active participant in your coding process.
• These features reinforce the idea that the future of software development lies in enhanced collaboration between human ingenuity and machine intelligence.
• As the ecosystem grows, we can expect even more integrations that further blur the boundaries between coding, debugging, and project management.

Concluding Thoughts​

This latest announcement is a clear signal that the world of coding is evolving—and fast. With Agent Mode now available to all VS Code users, combined with the added muscle of MCP support and versatile multi-model features, developers are empowered to take on challenges with unprecedented ease and efficiency. The innovative approach to premium requests further ensures that whether you’re an independent developer or part of a large enterprise, you have access to the right tools at the right time.
As Microsoft celebrates 50 years of innovation—a journey that began with BASIC and MS-DOS and continued through revolutionary platforms like the .NET Framework and VS Code—the future looks brighter than ever for developers around the globe. GitHub Copilot’s new capabilities are not merely incremental updates; they represent a holistic rethinking of how we interact with code, enabling an era where creativity is unleashed and routine tasks are automated.
In a world where every line of code matters, these tools are designed to help you write smarter, think faster, and build better. So, whether you’re diving into a new project or looking to optimize your existing workflow, the time to embrace these changes is now. After all, with Microsoft and GitHub pushing the boundaries of what’s possible, the next generation of technology is already in your hands.

Final Recap:​

• Agent Mode in VS Code now automates complex, multi-file coding tasks.
• Multi-model support, including leading AI models from Anthropic, Google, and OpenAI, offers a tailored approach to diverse coding challenges.
• The introduction of premium model requests and the Copilot Pro+ plan ensures that advanced tools are accessible to all types of developers.
• MCP support bridges the gap between your development environment and the AI assistant, opening up a new realm of contextual intelligence.
• Tools like the Copilot code review agent and next edit suggestions further streamline the coding process, allowing you to focus on innovation.

Embrace the future of coding with GitHub Copilot’s cutting-edge updates, and get ready to experience a new era of intelligent, context-aware development. Happy coding—and happy 50th, Microsoft!

---

Source: Microsoft Azure Vibe coding with GitHub Copilot: Agent mode and MCP support rolling out to all VS Code users | Microsoft Azure Blog