And that's why I described making a truly WYSIWYG HTML editor as a fair bit of work. Not at all impossible; I've made several site-specific ones over the years. But even site-specific tools will take some work.
HtmlText isn't at all HTML, as Hermann noted. Indeed, it was never intended to be. The purpose of HtmlText is to provide a plaintext representation of the contents of a LiveCode field, sufficiently complete to reproduce the field's content with complete fidelity.
But that's all. The name merely orients the user to the structure of that field content representation, and is not meant to imply a direct correlation with true HTML. For this reason the core devs once considered renaming the property "XmlText", but ultimately that choice would have been only incrementally better, still encumbered with expectations beyond its designed purpose, and would introduce compatibility issues making it a non-starter. So "htmlText" remains the name, despite the occasional misunderstanding as to its role.
That said, because HtmlText is machine generated it's predictably consistent, lending itself well to the massaging needed to transform into a form that can be rendered well in even modern browsers.
But here we soberly face the distinction between "possible" and "desirable": htmlText puts all formatting and styling inline, while there are many good reasons to separate those with CSS.
Given the consistency of htmlText, it's possible to build a system that uses style definitions which are applied to text runs. But the means by which this is done within LC will be different from what's needed in the browser, so any generator one would craft for that would require some thought about how that's handled and how the UI presents that to the user.
All of this is doable, but as Hermann reminds us, only if you're content with the subset of web presentation options supported in the LiveCode field object; other object types like images, buttons, menus, and vector graphics each require their own translation strategy.
To be fair, with widgets and custom controls, and a bit of scripting, even those need not be show stoppers, but it does mean more work for things like tables, many kinds of lists, images, etc.
Perhaps the biggest challenge is the difference between just about any app creation tool's object placement logic and the implicit flow of element rendering in HTML. App tools tend to use absolute coordinates, while HTML flows elements to avoid overlap.
True, one can generate the resulting elements with absolute coordinates to mimic how apps like LC work, but is that desirable in a web page where users are accustomed to content reflowing to fit a resizable window?
This requires the developer of such a tool to define a layout framework to translate the positioning we do in script to the corresponding CSS attributes.
Even better (a system I've been working on in my spare time) would be to craft a layout framework for LC which supports HTML-like flow. Such a framework not only benefits the generation of responsive web layouts, but also simplifies work for native apps that will never be have HTML generated from them. But I can say with the confidence of personal experience, this is not as straightforward as it might seem when you first get the idea over a beer with a friend.
In short, some parts of a "WYSIWYG HTML editor" are easier than others. As we learn more about the OP's specific needs we can provide useful guidance.