Weekly GitHub Copilot Roundup: Models, MCP, and Agents

by TechHub

GitHub Copilot delivered updated features this week, introducing premium model options, customizable workflows with Model Context Protocol (MCP), and new code editing capabilities to support more personalized AI-assisted coding. GitHub refined controls, extended automation, and consolidated context management in response to user feedback. Copilot’s transition from legacy environments to new collaboration solutions is ongoing, with organizations gaining more oversight for policy and spending, as illustrated by new case studies.

GitHub Copilot: New Models, Automation Workflows, and Core Feature Updates

Gemini 2.5 Pro is now available to all Copilot Pro/Pro+, Business, and Enterprise users. GitHub continues to support flexible, multi-model work with Gemini, OpenAI’s latest models, and more, enabling users to choose the AI that fits their work best. Gemini is selectable in Copilot Chat across major IDEs including VS Code, Visual Studio, JetBrains, Xcode, and Eclipse, and organization administrators can set Gemini as the default, following similar GPT-5 organization rollouts. Copilot provides commit message suggestions in the GitHub web editor for all users, moving automation beyond code writing to workflow documentation. Copilot Spaces improvements add precision in linking to repositories, directories, files, and branches, further supporting collaboration. These updates expand on earlier context management updates, making Spaces more practical for day-to-day team work. Workflow automation receives an upgrade with the Copilot coding agent and the Agents panel, available to all Copilot paid users. Developers can now assign tasks, track status, and manage pull request drafts directly on GitHub.com, shifting Copilot’s capability from code suggestion toward comprehensive project automation.

Copilot in Visual Studio and .NET Ecosystem: Diagnostics, Controls, and Integration

Visual Studio continues to improve Copilot for .NET with expanded debugging and smarter context support. The Copilot Diagnostics tool now brings AI-powered debugging—with automated breakpoints, LINQ analysis, and focused profiling for CPU and memory. Aspire 9.3 now includes Copilot for live troubleshooting and log review, following last week's integration of Copilot into cloud-native .NET applications. Refined Copilot controls in Visual Studio respond directly to developer feedback. Updated delay settings for completions, selective suggestion handling, and keyboard shortcuts help developers maintain control over automated code suggestions. Model Context Protocol (MCP) integration in Visual Studio has reached general availability. Using .mcp.json files, developers can now connect to custom or community MCP servers, simplifying adoption of secure context automation.

Model Context Protocol (MCP): Advanced Extensibility and Custom AI Workflows

Building on prior enterprise automation news, this week introduces detailed guides and general availability for custom MCP servers. With MCP, Copilot can connect to internal APIs and business applications, supporting tailored automation. Real examples—such as a TypeScript server for turn-based games or automated class naming—demonstrate how teams can use MCP to apply domain-specific automation within their preferred IDEs. With full support for MCP in Visual Studio, organizations can better manage secure orchestration and contextual workflows, especially where process control is essential.

Copilot Spaces and Context Management: Transitioning from Knowledge Bases to Spaces

Copilot Spaces is scheduled to fully replace the former knowledge base system by September 2025. This change consolidates Copilot’s context features into a unified platform. Spaces now support linking to code, documentation, images, issues, pull requests, and more, allowing for comprehensive context and information sharing. Users should move their content using migration tools before retirement. This change makes Spaces the central location for team-oriented, context-driven AI workflows.

Everyday Productivity, Workflow Automation, and Developer Experience

New resources this week emphasize Copilot’s impact on productivity within .NET and workflows involving multiple files and TDD. Guides show how Copilot aids in tracing dependencies, maintaining cross-file consistency, and organizing large projects. The system now provides support for TDD, prototyping, and code scaffolding by improving its understanding of file relationships. Recent case studies, such as Bank Galicia, demonstrate that Copilot is increasingly essential for teams seeking more reliable and faster delivery, evolving from small pilots to standard practice.

Copilot Customization: Bring Your Own Models and VS Code Personalization

Visual Studio Chat now lets developers integrate third-party models (such as OpenAI, Anthropic, or Gemini) by providing their API keys and selecting from available models, currently for chat-based use only. This builds on Copilot’s existing support for a growing set of models, including GPT-5, Claude, and Gemini. Copilot plans further flexibility for model selection. Recent guides for VS Code Joyride and Copilot demonstrate workflow personalization. Developers use Copilot within scripts and personalized tools, reflecting a wider trend of individualized automation routines.

Other GitHub Copilot News

New for Business and Enterprise customers, Copilot now features a request overage policy. This lets administrators set usage boundaries or enable overage pricing to control expenses—a capability needed as Copilot adoption grows in organizations. The announcement of GitHub Universe 2025 highlights technical sessions on Copilot, Actions, security, and community learning. As with past conferences, these sessions aim to help teams optimize their use of GitHub’s AI and collaboration tools at scale.