Ah, my bad! I know what you mean.
We do some kind of an hack to bundle images into Markdown files, using data-url on Rich Text exports.
I don't know about this really, it's not very practical either. I just loaded a 3MB image into the document and now I have a 2 millions characters url and can't properly edit it anymore.
From my side a .zip archive containing:
- the
.md document
- an
/img folder with the images inside
...would be much cleaner.