That's what my Meta-Machine Code is, as well. Still, perhaps others will be interested and I have recognized its potential value for pedagogy.
I avoid this issue in part by splitting the program into libraries where I found reasonable and then making those libraries rather comprehensive and well documented. So, only the MMC itself is in a poor state and relatively poorly documented; this is largely in part due a poor design aspect I pursued for a time, though.
I created a poor machine code as the customization language and much of the tool is written with it. Any reimplementation will lack this, in part because it's poor and in part because the current CHIP-8 targeting doesn't even need customizability.
I'm considering starting on the documentation for the tool already. I'm planning a book that is no more than one hundred pages. I suppose, ideally, it would be closer to, say, fifty, but I'll see. In any case, the tool is well-defined enough for me to begin work on this; I'd want to have my own system in place for book writing, though.
I've unfortunately noticed some queer bugs with this program, but I simply intend to have the semantics down and bugs gone with the reimplementation I'm planning. If you're the only ever intended user, this doesn't matter, though. I have programs I don't distribute that have some flaws or lack of functionality I simply work around.
My advice is to pick a single project and ruthlessly pursue only that project, then. My tool and its derived libraries were my only focus for quite a while, although I'm now designing or working on other libraries and programs I want to write, in particular things involving database backups, Gopher, and other things that interest me. So, I have a level of self control and motivation that lets me focus on something and then give some other things my attention when I'm fatigued, without failing to later return to it.
If you're struggling with just that, though, then obsess and focus on your main project. I have some much larger ideas that are still only vague ideas. What I did was order them sequentially and this tool was the first, as I intend to use it to build the others. Try that.