Feature
Keyword filters miss disguised rule-breaking and flag innocent messages that happen to contain the wrong word. AI AutoMod reads what a message actually means, escalates fairly, and leaves a paper trail your human moderators can review.
Enable it and EzraAI reads incoming messages for context — tone, intent, disguised phrasing — rather than matching against a static banned-word list. That means it can catch things a keyword filter would miss, and it's less likely to flag a message that just happens to contain a sensitive word used innocently.
Violations aren't treated as one-size-fits-all. EzraAI tracks a strike ladder per user, so a first slip-up gets a lighter response than a pattern of repeated violations. Every flag can be routed to a log channel with /set-log, so moderators see what happened and why — AI AutoMod supports human moderators, it doesn't replace them.
Beyond moderation, EzraAI includes a ticket system for handling member requests and reports in an organized channel, with FAQ auto-replies to cut down on repetitive questions.
| Command | What it does |
|---|---|
| /automod | Configure AI AutoMod strictness and exemptions for a server (admin) |
| /set-log | Set the channel where moderation flags are logged (admin) |
| /report | Report a message or user |
| /set-reports | Configure where reports are routed (admin) |
| /smart-slowmode | Automatically adjust channel slowmode based on activity (admin) |
| /ticket-setup | Set up the support ticket system (admin) |
| /ticket | Open a new support ticket |
| /ticket-close | Close an open ticket |
| /ticket-reopen | Reopen a closed ticket |
| /ticket-faq | Configure auto-reply FAQ entries for tickets (admin) |
For the full rules on what's prohibited and how enforcement works, see the Safety & Moderation Policy.
It reads what a message means instead of matching a fixed word list, catching disguised rule-breaking without flagging innocent uses of a sensitive word.
Violations escalate through a strike ladder, and flags can be routed to a log channel for human moderators to review.
Yes, through the support server, on a best-effort basis.
