Microsoft’s Visual Studio Code has hit a new milestone with its April 2025 update, version 1.100, introducing a significant evolution in its AI-powered Chat capabilities and overall editor experience. Developers now gain unprecedented control over AI interactions through new customizable instruction files (`.instructions.md`) and reusable prompt files (`.prompt.md`), allowing for tailored coding assistance aligned with specific project needs and team practices.
The update also begins the gradual rollout of OpenAI’s GPT-4.1 as the default base model for Chat, promising more sophisticated coding suggestions and understanding. The enhancements aim to make AI an even more integral and efficient co-developer.
The focus on AI extends to new Chat tools designed to streamline common developer workflows. Users can now employ the `#githubRepo` tool to search for code examples or implementations within any accessible GitHub repository directly from the chat interface, without needing to clone or open the repository locally.
Additionally, the `#extensions` tool simplifies the discovery and installation of add-ons from the Visual Studio Code Marketplace. These additions, coupled with performance improvements like faster agent mode edits and quicker responses to repeated chat requests due to conversation summarization and prompt caching, underscore Microsoft’s commitment to a more responsive and intelligent coding environment.
This version builds on features progressively introduced in recent updates, such as Agent Mode which became available in VS Code Stable in March’s version 1.99, and Next Edit Suggestions (NES) which reached general availability that same month. Version 1.100 further refines NES with a new, faster model.
The introduction of distinct prompt and instruction files in this latest release offers a more structured approach to AI guidance. According to the official release notes, instruction files provide context and guidelines, like coding styles, which can be applied automatically to relevant files or attached manually. Prompt files, on the other hand, encapsulate complete, reusable chat requests, including mode and tools, runnable via slash commands or an editor play button. The Visual Studio Code Release Notes state, “You can tailor your AI experience in VS Code to your specific coding practices and technology stack by using Markdown-based instructions and prompt files.”
Enhanced AI Interaction and Editor Experience
Visual Studio Code 1.100 deepens AI integration by allowing developers to tailor their experience with Markdown-based instruction and prompt files. Instruction files (`.instructions.md`) enable the definition of common guidelines and context, such as code style rules or preferred frameworks, which can be automatically applied to specified files or manually attached to chat requests.
Prompt files (`.prompt.md`) facilitate the creation of reusable, standalone chat requests, complete with mode (ask, edit, or agent) and tool specifications, making it easier to automate common tasks like component creation or security reviews. The Visual Studio Code team details that these files can be stored in user data folders for cross-machine synchronization or within workspaces.
Beyond chat customization, version 1.100 introduces several editor and workflow enhancements. New floating window modes – ‘compact’ and ‘always-on-top’ – provide more versatile multi-window setups. The ‘compact’ mode minimizes UI elements in floating windows for a cleaner look, while ‘always-on-top’ keeps a selected window persistently visible. These features are particularly useful for keeping a Chat window accessible while working in the main editor.
Source control sees the addition of quick diff editor decorations for staged changes, allowing developers to view these changes directly within the editor. Security continues to be a focus, with mandatory extension signature verification now extended to Linux platforms (excluding ARM32 builds due to a known issue #248308), ensuring all extensions are validated before installation. For extensions flagged as malicious, VS Code will now provide ‘Learn More’ links for additional context.
For developers working with Python, the Pylance extension now offers an interactive color swatch in the editor and an experimental AI Code Action for converting string concatenations to f-strings or `format()`. The Python Environments extension introduces a Quick Create command for virtual environments and new chat tools for environment information and package installation.
The GitHub Pull Requests and Issues extension also sees improvements, including the ability to ask chat questions about the active pull request. Extension authors benefit from the finalization of the Text Encodings API and initial ESM support for NodeJS-based extensions.