Okay, gotcha! So the statblock has "Tiny" or "Diminutive" and that screws up the statblock2token macro.
It looks like the S2T macro uses the size from the beginning of the statblock (Diminutive, for example) instead of the
Space field later in the block. It should be using Space, since that's the footprint of the creature on the 5-ft grid.
However, due to the way the v1.8 S2T macro works this isn't easy to do. I haven't looked at v2.0 of the S2T macro so it might be better.
I really had planned to rewrite the S2T macro when Imper1um came along and decided to do so! (It had only been on my todo list for about 8 months!

) I was thinking the best approach was to
tokenize the entire statblock and then pick out the information between tokens. This would work for the simple cases, but to truly do this job correctly would require a full-blown parser. And that's not worth the effort.
The good news is that
others have done some of the work of writing parsers in JavaScript, so at some point in the future this particular task will be much simpler. Although it will still require someone who understands both parsers and statblocks, and in particular how statblocks might vary.
Actually, it looks like
this one runs in a web browser window and can be fed the grammar prior to parsing. This is likely the easiest way to play around, although MT will need something slightly different.
Oh well...