Post

💯 v1.100 of VS Code is here! And we’ve got some great updates for you, like: - Smarter chat responses with new tools - Improved multi-window support - Image and Streamable HTTP support for MCP servers …and so much more. aka.ms/VSCodeRelease Here are some of the highlights… 🧵

Visual Studio Code

@vscode.dev

8 May 2025

There are two new modes that floating windows can have: - Compact, which hides certain UI elements. - Always-on-top, which stays on top of all other windows Pro tip: use it with the Chat view.

Visual Studio Code

@vscode.dev

8 May 2025

You can now use the built-in Simple Browser to attach UI elements as context to chat, allowing you to ask about specific elements of a web page.

Visual Studio Code

@vscode.dev

8 May 2025

Chat now has an LLM tool to run notebook cells, which allows the agent to perform updates based on cell run results or perform its own data exploration.

Visual Studio Code

@vscode.dev

Prompt files are useful for creating reusable chat requests for common tasks and include the prompt text, chat mode, and tools to use. You can open the prompt file and press the "Play" button to run the prompt. You can also use the Chat: Run Prompt File command to run.

8/5/25, 5:18 pm0 reposts2 likes

Visual Studio Code

@vscode.dev

8 May 2025

Or you can run a prompt file by typing / in the chat input field, followed by the prompt name.

Visual Studio Code

@vscode.dev

8 May 2025

This release adds support for the new Streamable HTTP transport for MCP servers. Streamable HTTP servers are configured just like existing SSE servers, and our implementation is backwards-compatible with SSE servers.

Visual Studio Code

@vscode.dev

8 May 2025

We have enhanced the UI that shows MCP server tool input and output, and we have also added support for MCP’s new progress messages.

Visual Studio Code

@vscode.dev

8 May 2025

When your conversation gets long, or your context gets very large, you might see a “Summarize conversation history” message in your agent mode session. This is because we’ve made some changes to optimize for prompt caching in an effort to speed up model responses.