How the Hub Is Moderated: Upload Checks, Review, Reporting
"Strictly SFW" is easy to claim on a website. To make it more than a claim, every publish on the Hub runs through a multi-step check — automated, with manual follow-up review for anything borderline.
What happens when you click "Publish"
Images run through an image tagger that rates content; character text — persona, greeting, example dialogue — runs through a text classifier that also checks written content for sexual or minor-related depictions. The whole pipeline runs synchronously before anything becomes visible: you get the result up front, not a rejection later. Important: this only applies to Hub uploads. Anything that stays local in the app is never scanned.
Field-by-field feedback instead of a black box
Instead of a blanket "rejected," you get a report per field: persona checked, greeting flagged, one example dialogue disputed — with the quoted passage or affected image, a category, and a link straight to the right spot in the editor. Borderline cases go automatically into a manual review queue instead of an automatic rejection. After a revision you can re-check everything; an unchanged card isn't checked twice.
The one hard line
There's exactly one thing with no feedback loop: depictions of minors. A violation results in rejection with a deliberately cryptic error code instead of a detailed explanation — otherwise the filter logic could be probed and gamed. Repeated attempts lead to an account ban. This rule is non-negotiable and applies regardless of whether it's fiction.
If something still slips through
Every published card can be reported — even without an account, even anonymously. The report form asks for a reason (depiction of minors, sexual content, copyright, other) and a short description. Reports go exclusively to the moderation team and are handled confidentially; knowingly filing a false report can have consequences for your own account.
The Hub checks what you publish — nothing in the app is ever scanned locally.