Four independent guards for links, images, and voice content.
Each is a page of its own with its own settings, actions, and exemptions: link guard, image spam, image scan, and voice scan. You enable only what you need without it affecting the rest.
Four fully independent guards, each a page with its own settings, actions, and exemptions: link guard, image spam, image scan, and voice scan. You enable only what you need.
Link guard has three modes and the difference between them is large. Smart blocks only scam and phishing links and lets normal conversation through, which suits most servers. Whitelist blocks every link except from domains you allow. Strict blocks all links. Alongside them sit domain lists, invite code lists, and a test box where you paste a message and see the verdict before applying anything.
Image scan has two engines: a free local one with no key that catches inappropriate content and reads text inside the image to spot scam screenshots, and a more accurate cloud engine you can layer on top if you need it. You set the local sensitivity as a percentage, plus the cloud strictness and its minimum confidence. It catches nudity, gore, insults written inside the image, and scans GIFs if you want.
Voice scan watches audio content: insults, threats, sexual content, and racism, with a strictness level and a minimum confidence before any action. Image spam is the simplest: a number of images within seconds, a sticker limit, and blocking duplicate images.
All four share the same actions: delete and warn, or a timeout or jail, with repeat-offender escalation, a warning message with its variables, and exempt roles and channels.