This is definitely something you can do within ProseMirror, but it’d be complex as one would expect. Let me try to give some pointers.
In my editor I would like to give users the ability to add a translation for each word they type using a custom markup syntax like this: word(translation) or like this: [word]words with spaces/word.
The translation itself will need to be stored somehow, likely as a data attribute on the element.
The first question to think about is whether you would want to use a node or a mark. Both can support storing data within the attrs field.
However, where this gets complicated is that I want the markup itself to disappear when the user hits space after completing it, so e.g. “word(translation)” just becomes “word” maybe with a custom css class to indicate the word has a translation.
Markup to disappear (or to appear) when the user hits space can be accomplished with a mix of decorations and perhaps an inputrule to create the node or mark to store the translation attribute.
Then, whenever the user moves their caret back into the word I want the markup to reappear, e.g. “word” becomes “word(translation)”. It’s this last part I just can’t figure out in Slate js.
This would involve checking on each editor update whether the current selection intersects with a word node or mark. If it does, then you recompute the decorations (or the nodeview) to show (translation).
Overall, I would try grokking the source code of prosemirror-math because they do something similar with the cursor selection to switch between editing latex source and latex render. Best of luck!