List of steps to reproduce:
What happens?:
Page content model: wikitext
What should have happened instead?:
Page content model: JSON
Software version (if not a Wikimedia wiki), browser information, screenshots, other information, etc.:
Probably requires an adminbot.
In own userspace, a JSON can be created. Once created, it should be possible to move it to Project: while retaining the JSON PageContentModel. (this works on enwiki, not tested on meta) This could maybe be used as a workaround.
A good alternative however would be to enable Data: on meta. Data: is also available on mediawikiwiki, but it's not really appropriate to put Web2Cit there.
Before someone else brings this up: I know Commons has a Data: namespace. From a technical standpoint it's better suited for the task. I urge everyone involved not to use it. Users are way more likely to get blocked on Commons (due to copyright violations because copyright is very complicated) which would also exclude them from Web2Cit if the data was moved there. There's also a severe problem with admin integrity. For the record, fabricating death threats and hiding behind "sorry, confidential" and "I don't wanna talk about this" (the whole CheckUser team was involved in that, not a lone bad apple) when asked for evidence is not OK. The CheckUser team ultimately got away with this. Commons is not safe, Web2Cit data should not be moved there.