Playable demos & bots
jev-content-mod asks Jev about harm, category and clarity, then applies separate code rules for approval, hidden review or removal.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
Jev scores harm, chooses a category and answers clarity questions; code applies approve, shadow or remove rules. The example requires 0.85 clarity for removal and 0.5 for approval. These are hand-set policy values with no reported evaluation, not universal moderation defaults.
The three-way policy keeps unclear, harmful-looking posts in a review band instead of automatically deleting them. That matters when a wrong deletion is costly. Choose cutoffs for your own community and test known cases; the sample's numbers represent the author's tradeoff, not a measured universal policy.