ensardev

Actually, my project supports this, and I can package for macOS using Tauri, so with a minor adjustment, I can generate a build for macOS. However, the major hurdle I face is Apple’s approval process. Apple Gatekeeper has stricter rules than Windows’ SmartScreen, which is why the build I create must be a signed and approved version. Since I don’t yet have an Apple developer account, I haven’t felt the need to create a build for the macOS environment. However, looking ahead, I plan to set up my primary development environment for macOS and prioritize Apple (while also covering Windows and Linux). So this might be possible in the near future.

“Actually, you're right—as you said, WebView usage isn't included in that figure. The 10 MB you see is actually just the footprint of the Rust binary. Still, I have to say that Tauri has turned out to be a truly excellent tool in this regard. Compared to Electron, since the WebView2 runtime is shared with the operating system (each package doesn't bundle its own Chromium), the system overhead remains very light.”

And yes, you’re getting the SmartScreen warning because this build hasn’t been signed yet. This is generally a bit of a hassle for individual developers. However, I’m looking into this issue.

I didn't take it as criticism :) Thanks.

From what I can tell, CyberChef is a more comprehensive and well-established project. Of course, there are many projects like this. I wanted to work on a project like this because I thought it would be a good fit for Vibecoding. It would also be a good opportunity to try out Tauri and other tools.

I definitely agree with what you said about Tauri. Tauri’s performance surprised me.

As for what you said, to be honest, this project was just a starting point for me in terms of getting to know Tauri. So, in the context of this project, I’d say Tauri was more of a shell designed to deliver the application natively, rather than a tool where I fully utilized all its capabilities. In that regard, I didn’t encounter the kind of issue you mentioned. I mostly noticed that the AI initially struggled with window control and couldn’t grasp the situation. In short, when it came to ACL, I had to identify the problem and guide it. The AI had written reasonable window control code but had overlooked what to do regarding permissions.