Weekly GitHub Copilot Roundup: Multi-Model IDEs and Agent Skills

by TechHub

This week, GitHub Copilot introduced broader model access in leading IDEs and enabled more tailored workflow customization. These updates span both real-world developer scenarios and educational resources, highlighting Copilot's wider support in conventional programming environments and creative use cases.

With continued support for multiple agent types and improved model integration, the Gemini 3 Flash model for Copilot Chat is now available in public preview. Visual Studio (17.14.12+ and 18.1.0+), JetBrains IDEs (1.5.62+), Xcode (0.46.0+), Eclipse (0.14.0+), github.com, and GitHub Mobile now let users select Gemini 3 Flash via the chat model picker. This makes Copilot’s multi-model support more accessible in standard developer environments. Modes including chat, ask, edit, and agent are available in Visual Studio and JetBrains IDEs, while Xcode and Eclipse provide ask and agent options for custom workflows.

Copilot Customization and New Workflows in Visual Studio Code

Extending the focus on project-specific workflows in VS Code, the December 2025 (v1.108) release adds Agent Skills for GitHub Copilot. Developers can save specialized knowledge in SKILL.md files within .github/skills and load them during Copilot chat sessions, supporting individualized guidance and project context. Updates also bring improved navigation, session management, keyboard and git workflow support, and better debugging and accessibility. Copilot plugin development and issue handling gains new tooling for code quality and maintainability in both enterprise and open source environments.

Tutorials: Real-World, Agentic, and Spec-Driven Copilot Workflows

New tutorials demonstrate Copilot usage across hands-on projects and agent-driven approaches. The Space Invaders walkthrough shows Copilot integrating with Figma, implementing advanced gameplay mechanics, Playwright code reviews, and structured feature specs, supporting consistent team practices and automation with MCP. Azure MCP and CLI deployment strategies build on recent teamwork themes using intelligent agents. A Copilot lab for Visual Studio Code explores agent mode, covering MCP server connections, custom agent development, and automated background tasks. These learning resources help developers move from basic completion to orchestrated multi-agent workflows, continuing the shift toward scalable, enterprise-grade Copilot implementations.

Analyses: Copilot’s Impact on Frameworks, Open Source, and Coding Standards

Analysis this week looks at how Copilot influences technology choices and engineering approaches. Articles discuss Copilot’s role in supporting React’s ongoing use, building on past themes about AI and statically typed languages and developer efficiency. An interview with Homebrew maintainer Mike McQuaid spotlights open source automation and scaling strategies involving Copilot. Emphasis is placed on bug and documentation management through automated AI processes previously discussed in agentic and MCP concepts. A case for moving beyond informal code styles to systematic, specification-led workflow development supports a disciplined approach to software engineering as AI becomes more integral to project delivery.

Other GitHub Copilot News

This week’s case study highlights Microsoft Copilot helping Deaf creators. Min-ji So’s Webtoon Storyboard Assistant uses conversational AI for grammar suggestions and creative support, further demonstrating Copilot's accessibility and flexibility across industry and creative fields. These examples reinforce Copilot's growing role in enabling new workflows and fostering broad user impact.