djlewald

At the start of this year, I vibe-coded this email client to replace Apple Mail for my daily use. The original goal was something native and lightweight, but it quickly morphed into tinkering with the idea of viewing my inboxes as feeds (and organizing them that way). It's still lightweight, but now includes a tagging system that automatically categorizes emails so they can be displayed in a multi-feed format.

Like many others, something I built to learn what frontier models could do has become something I use daily. After months of use and refinements, I figured it's time to let everyone else in on it (haha).

For now, the project's scope is still narrow. MacOS only, Gmail only. The organization is purely client-side. Other than sending email, nothing gets reported back to the email server. I have set it up in a way that it could support other email services, namely outlook. And given the tooling, it could be deployed to other platforms (Windows a/o Linux). Personally, I do all my emailing from a Mac, hence why I haven't lol. But if others wanted to light that torch I'd be happy to support them!

From what I understand, there's no standard implementation of the standard. So each implementation is unique in some ways. Therefore a "unified" experience is impossible.

https://youtu.be/k3bUvZf89k8?t=697

edit: I also like to buy merch from OSS affiliated companies. e.g. mozilla hoodie, Neo4j shirt etc.

Edit: I now see their producthunt page. It didn't pop up while searching /shrug

The method chaining feels good as long as you check the return type of the function your calling. Most of the time it returns the element you just asserted against. And when you're trying to drill down into deeper sections of the DOM you can speed up tests by calling .within from a higher level element to keep the search to a specific part of the tree.

The individual steps with their state available for viewing in the sidebar is great as others have mentioned. The test doesn't run at light speed, so locally you can actually VIEW what is happening. Plus when it does fail there's typically a long timeout (~40 seconds) before the test actually fails. So it's very noticeable when your test hits a snag.

Also helper functions are not hard at all, and we have a few useful ones we use in almost every section. I have more I could say, but I'm not going to write a huge raving post. Just wanted to try to post some positive contrarian perspective.