I used fable to build sync for zen
tbh, this was just for fun.
this is one of the biggest problems I have with zen. it's a fantastic browser, but it doesn't have sync (like pinned tabs, folders, essentials all of this fancy stuff). so i was like, "hey, fable has been around for a bit, i have access to it through claude pro, why not see if i can get this feature?" it's supposed to be the most powerful coding model and all that.
and tbh, I actually thought it would do it.
man... it is rough.
to be fair, I wasn't asking it to build everything from scratch. there are already ongoing prs tackling this exact problem, and the people working on them are doing a fantastic job. I honestly don't think fable would've gotten this far without their base work.
you can find all of my experiments, commits, and the chaos here:
https://github.com/ToYoNiX/zen-browser-desktop
you can even build it if you want. it runs, it functions, but it is buggy.
like, I'm still not even sure how tab ordering is handled anymore because during sync tabs can apparently just teleport. if you close a tab across all your devices, it'll sometimes just come back out of nowhere. live sync is a mess. yes, it mostly works, and maybe with a bit more prompting and a lot more token burning I could get it over the finish line, but... come on.
at this point, I'm convinced that vibe coding is mostly good for either starting something completely new or generating code when you already understand the underlying logic and just want to write it faster.
other than that, I'd much rather implement it myself and actually understand what's going on. after trying this, I have way more appreciation for the people working on the real sync implementation. this stuff is a lot harder than it looks, and I'd rather wait for something solid than pretend ai can brute-force its way through all the edge cases.