I am the developer of the Paste and Replace plugin. Thanks for you feedback I will try to look for a way to keep the override between components.

Our new design language is called Fluent, but when I built this 2 years ago the system we used was Fabric.

No worries, and yeah I did this mostly for my own interest and learning. This project was actually done 2 years ago, and back then there wasn't a unified team building designer tools like icon libraries. Now we have an official internal icon library built by a dedicated tooling team, but many designers still prefer my tool because of some features that are not available in the official one (e.g. tagging, download svg, upload svg, double click to copy etc.)

I know right! By the way, thanks for checking out this article. (I'm the Microsoft designer who built the Icon Library) It was a great opportunity for me to share my little side project to the world!

Exactly. Too many times research showed improved metrics between two designs with multiple varied factors, while may be only one or two things in the new design actually contributed to the success. It's too simplistic to conclude everything in the new design works better without drilling into what individual elements actually made it better.

Works just fine with Zeplin and other redlining tools. There's no need to have weird sublayers, since everything is done on the layer name. For things that don't have a colored background, I sometimes just add a background rectangle layer with no fill. This actually helps redlining because it shows the actual 'div' box around the element instead of what's only visible to the eye.

For example, with Paddy you can set the paddings of a button so when the label text changes the background would automatically updates its size, just like real CSS would. You can also set a persistent spacing between elements in a group so they are always say 8px apart. The combination of these two features makes for an extremely efficient workflow where you'll never have to nudge a single pixel every time you change the content. This is so powerful that it should really be built into one of these design tools.

I really like this concept, especially the interaction for case 2.

I think a better solution for single word correction is just always show the word to be replaced in blue, if user agrees he/she can tap the blue word to execute the replacement, rather than it being replaced automatically for the need of an undo mechanism.

And for multiple matched words, as in case 2, I think it may be better to highlight all the matched words and wait for the user to tap on one of them, since he/she may not have paid attention to the quick flashing words and missed which ones were tappable.

And if no correction is needed, the next message sent by the user will cancel out the highlighting, ending the interaction.