This would make it easier to store machine-readable data which can be used for storing index-level data such as:

• Things that are currently in fields of the template:
  • Proofread status
  • Transclusion status/date
• Things that would be awkward to store in this way (because they'd be hard to read out of Wikitext):
  • E.g. Work-specific snippets/character sets
  • Scan offsets for high res 3rd party page links T276437

This can then be accessed by API (for gadgets and bots) as well as via PHP for handing over to the Lua library.

I suppose that it would probably need Mediawiki:Proofreadpage index config to be able to select which fields go to JSON and which are passed onto the index template as they are now (the template can still read the JSON data though a module)