Microsoft has very specific constraints on what extensions can and can't do, it's not a free for all. They're actively defending their mote by allowing Copilot to do things in a way that extensions couldn't. That's why all the serious contenders make a fork, it's simply not possible to have the same integration otherwise.
Because it just searches Google and HN is indexed regularly, nothing really noteworthy. If you copy paste the same quote into Google you get the same thing.
Still it would be a lot wiser if all the forkers would do one 'AI-enabled' fork together that exposes all the extras that copilot gets. The barrier for testing would be much lower and all the extension makers would also jump onto the train. Likely MS would finally give in and make all the extras available for everyone. But all the fragmentation only helps MS.
I've had a Github Copilot subscription from work for 1yr+ and switch between the official Copilot and Roo/Kilo Code from time to time. The official Copilot extension has improved a lot in the last 3-6 months but I can't recall ever seeing Copilot do something that Roo/Kilo can't do, or am I missing something obvious?
The Copilot extension uses proposed APIs, meaning it's on an allowlist bundled with VS Code. Roo likely enables these early. The API can stay proposed for years before Microsoft opens it up to third party users.
They're all going to have quite divergent opinions on how to structure the fork and various other design decisions that would inevitably lead to forks of the fork again.
I think forking VS Code is probably the most sensible strategy and I think that will remain the case for many years. Really, I don't think it's changing until AI agents get so ridiculously good that you can vibe code an entire full-featured polished editor in one or a few sessions with an LLM. Then we'll be seeing lots of de novo editors.
The Claude Code extension on VS Code does very little (too little in my opinion). The integration level with agentic functionality provided by Antigravity goes much deeper in my 20 minutes or so of playing with it. The biggest value pieces I see is: Agent Manager window which provides a unified view of all my agents running across all my workspaces (!) where I can quickly approve or respond to followup questions and quickly brings me to the code in context for each agent, additionally, I can select a piece of code and comment on it inline and that comment gets sent to the correct, active agent. These two things alone are items which I have been looking for... Too bad I only have approval to use Claude Code at work. This looks promising.
Well, you are entitled to your opinion but many people would disagree with you, and that's the crux of the issue, everyone has their own conflicting views on what the UX should be, hence all the forks.
I don't even know what the Claude Code extension does in vscode. I have it installed but hell if I know what it's doing. I run Claude in one of vscode's terminals, and do everything through there. I do see (sometimes) diffs pop up in the IDE, I guess that's the extent of this integration.
