VS Code: Chat Gets Claude Rules and Context Smarts

Transcript

Hey there, VS Code community! Welcome back to another episode where we dive into the latest and greatest happening in everyone's favorite editor. I'm your host, and wow - do we have an exciting episode for you today. The team has been absolutely crushing it with some major chat improvements that are going to make your AI-powered development experience so much smoother.

Let's jump right into the big story of the day - and that's all about making chat smarter and more intuitive. The star of today's show has to be the Claude rules support that Martin Aeschlimann shipped. This is huge! With over 500 lines of new code across 18 files, we now have proper support for Claude rules in the chat system. What this means for you is better, more consistent AI interactions that follow the specific guidelines you set up. It's like giving your AI assistant a proper rulebook to follow.

But the chat improvements don't stop there. Megan Rogge delivered something I'm personally excited about - contextual tips. These aren't just any tips - they're smart tips that actually pay attention to how you're using VS Code. If you've already mastered agent mode or instruction files, the tips won't keep nagging you about them. It's that kind of thoughtful UX design that makes VS Code feel like it's really paying attention to how you work.

And speaking of Megan's great work, she also added context menus to those chat tips. Now you can right-click and either dismiss individual tips or turn them off entirely. Sometimes it's the little quality-of-life improvements that make the biggest difference in your daily workflow.

Behind the scenes, Matt Bierner has been doing some serious architectural heavy lifting. He's moving the chat sessions to a push-based controller model, which might sound technical, but what it really means is fewer unnecessary refreshes and better performance. It's the kind of foundation work that makes everything feel snappier without you even noticing.

The terminal got some love too, with Daniel updating xterm to version 6.1.0-beta.152. These regular updates keep the terminal experience smooth and bug-free. And there's a nice addition for security-conscious folks - the terminal sandbox now supports trusted domains, giving you more control over network access.

One of my favorite fixes came from the accessibility front. There was an issue where the chat progress sound wasn't working properly for queued requests, and Megan knocked that right out. It's these attention-to-detail moments that show how much the team cares about making VS Code accessible to everyone.

Paul added some powerful new capabilities to tool hooks with support for continue, stopReason, and systemMessage. If you're building extensions that integrate with the chat system, these additions give you much more control over the conversation flow.

The editor itself got a neat enhancement with a new "offWhenInlineCompletions" option for quick suggestions. This is perfect if you want suggestions to step back when you're already getting inline completions from Copilot or other AI tools. No more suggestion overload!

There were also some nice telemetry improvements for remote tunnels, better theme CSS extraction for reusability, and various bug fixes that keep the experience polished.

Today's focus is all about embracing these chat improvements in your workflow. If you're using AI assistants in VS Code, take some time to explore the new Claude rules support - set up some guidelines that match how you like to work. Try out those contextual tips and see how they adapt to your usage patterns. And if you're building extensions, definitely check out the new tool hooks capabilities.

What I love most about this batch of changes is how they show VS Code evolving not just with new features, but with genuine intelligence about how developers actually work. The contextual tips, the smart suggestion management, the performance improvements - it all adds up to an editor that's not just powerful, but truly thoughtful.

That's a wrap on today's episode! Keep building amazing things, and remember - every small improvement in your tools compounds into bigger productivity wins over time. Until next time, happy coding!