GoKawiilEmDash adopts Cloudflare's Clef AI model to vet plugin listings
EmDash now runs every submission to its official plugin registry through Cloudflare's Clef decision model before publication. Clef evaluates listing text, links, icons and screenshots against roughly a dozen categories including phishing, impersonation, hate speech, scams and moderation manipulation, assigning a probability score to each. Listings scoring 0.45 or higher in any category are escalated to a human operator rather than being blocked automatically.
GoKawiil's interpretation of the reporting above, not reported fact.
Because the registry runs on AT Protocol, plugin authors publish signed releases directly from their own accounts, so EmDash cannot alter the underlying record but can still control what surfaces on its public catalog site. Automated screening lets EmDash preserve that open, decentralized publishing model while reducing the risk of scams or malicious plugins reaching users, without requiring manual review of every submission. Keeping a human in the loop for borderline cases suggests EmDash is trying to balance moderation speed against the risk of wrongly blocking legitimate plugins.
- Cloudflare's Clef model now screens every EmDash plugin listing before it appears in the catalog.
- Clef checks text, links, icons and screenshots across about a dozen abuse categories and outputs probability scores.
- Listings scoring 0.45 or above are flagged for human operator review rather than being auto-blocked.
Source: emdashcms.com — Scott Buscemi, 2026-10-07
Published there as: “EmDash uses Clef to moderate the plugin registry”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.