Hello @thomasmarwitz 🙂

That depends on how nbconvert works, how we can hook into it, or how much mkdocs-jupyter relies on it. Essentially, since mkdocs-jupyter is a MkDocs plugin, it definitely has access to the Python-Markdown Markdown instance created by MkDocs and used to convert Markdown to HTML. So it could use it to convert Markdown cells, and this md object would have the mkdocstrings extension (and by extension, no pun intended, the mkdocs-autorefs extension too) registered, so mkdocstrings instructions (:::) and autorefs cross-references ([title][object.path]) would work (as well as all other markup supported by extensions configured in mkdocs.yml). Maybe with a few limitations, or hacks to apply. Happy to help if you want to try going in that direction 🙂