Visual Studio Code continues to evolve with new features and extensions aimed at enhancing developer productivity, particularly concerning AI agent development. Recent updates to VS Code, including versions 1.108 and 1.109, introduce several improvements and new functionalities.
One of the groundbreaking features is the introduction of Agent Skills. This experimental feature allows developers to enhance GitHub Copilot by teaching it new capabilities using folders that contain instructions, scripts, and resources. Skills are stored in <code>.github/skills</code> directories and come with <code>SKILL.md</code> files. These skills are designed to load automatically when they are relevant to a developer’s request, thus enabling automation and context-aware workflows for tasks such as documentation writing, image manipulation, and web testing.
Furthermore, the Copilot Studio Extension has been launched to facilitate the building and management of AI agents directly within VS Code. This extension provides comprehensive support, including IntelliSense, language support, and authoring capabilities for AI agent components. It integrates seamlessly with standard Git workflows, allowing for effective version control, pull request management, and synchronization of changes with Copilot Studio for testing. The VS Code environment assures compatibility with various AI assistants like GitHub Copilot and Claude Code.
The recent updates also focus on terminal and tool improvements. Version 1.109 introduces support for the kitty keyboard protocol, <code>win32-input-mode</code>, and new SGR escape sequences to enhance text formatting. Additionally, there are updates for the Agent Sessions view, including better keyboard access and multi-session archiving, thus improving chat functionality and session import options.
New features extend to terminal IntelliSense defaults based on user feedback, prevention of shell history for terminal commands, and profile imports via drag-and-drop. API improvements have been made for quick input buttons, overflow menus, and MCP server management commands, which now include MCP Apps support for custom UI.
To enable the Agent Skills feature, developers must turn on the ‘use agent skills’ setting in VS Code and incorporate <code>skill.md</code> files into their GitHub directory. This open standard ensures that skills remain portable across compatible agents.
Overall, these updates and extensions in Visual Studio Code empower developers with more robust tools and workflows to efficiently manage AI agent development while aligning with established software development best practices.

Collections

VS Code versions 1.108 and 1.109 introduce Agent Skills, an experimental feature that lets developers extend GitHub Copilot with custom capabilities using instruction files stored in `.github/skills` directories. The new Copilot Studio Extension enables building and managing AI agents directly in VS Code with IntelliSense support and Git integration. Updates also include terminal improvements with kitty keyboard protocol support, enhanced Agent Sessions view with multi-session archiving, and API improvements for quick input buttons and MCP server management.

Introduction of Agent Skills in Visual Studio Code 1.108