A major new version of Pandoc has just been released. It features a major set of new features and lots of underlying code updates to simplify future development. A full, rather epic changelog is available here:
pandoc.org/releases.html#pandoc-2.0-29-oct-2017
I’ve been using the beta version of Pandoc 2 with Scrivener for many months without issue. Of note for Scrivener users, there are some nice new additions to the pandoc markdown format. The first are fenced divs (pandoc.org/MANUAL.html#extension-fenced_divs), which allow you to create attributes on block paragraphs; for example you can create an INFO or WARNING box:
::: warning :::
some text
:::
:::{.info}
some text
:::
If you need to pass through some raw text to a specific format there is now a more general raw_attribute, so if for example you want to pass Mediawiki markup directly you can do inline like so: Here is some '''text''' to play with, <sup>yes</sup>.{=mediawiki} and the same for blocks using code block syntax.
Filters are a really powerful way to transform text from Scrivener, and Pandoc 2 now allows you to use native Lua filters without installing anything else: pandoc.org/lua-filters.html
You can now directly create PDFs via HTML using PrinceXML or weasyprint. I use Prince which works wonderfully direct from Scrivener.
Improved iBook metadata, and quite a few improvements to EPUB2/3 generation in general.
Many many fixes to improve DOCX, HTML, LaTeX and many other outputs.