I know, from using a network inspect tool, that you can use tags/autocomplete.json to fetch automatic completion for partial tag names. The way I have been using it has been (crude example):

fetch(`https://e621.net/tags/autocomplete.json?search[name_matches]=${cleanedUserInput}`).then ...

This works fine, apart from the fact that it also searches for antecedent names. Of course, you can just filter them (if (tag.antecedent_name == null) ... ) , but that leads to an inconsistent amount of tags left.
You could also "manually" search tags by using the normal tags.json, however using search[name_matches] requires wildcards, which leads, again, to inconsistent results and search[fuzzy_name_matches] sometimes neglects letters completely.
My question is: are there any other search parameters apart from name_matches, like limiting the amount of results or the aforementioned skipping of antecedent results or this a hard-coded thing?

That endpoint is https://github.com/e621ng/autocompleted, and it only accepts that singular parameter