For this, I assume I would need to then parse the AST?
Yes and no. The AST obtained from Raku can be converted into a hash (of nested hashes), and that is easily converted into JSON. This JSON output can be imported in Mathematica and then converted into the tree form you showed. We can say that that is parsing, but it can be done with none-parsing operations. (E.g. replacement rules.)
In all fairness, if that form is of (sufficient) interest, then we can make:
- A dedicated notebook expression converter
- Another interpreter in the Raku package
- Another Markdown parser-interpreter in Mathematica/WL.
I think option 1 is best. (Except, if you really want to use the Markdown file as a "single origin truth.") Also, for option 1, Kuba's package "M2MD" might be a good start.