Skip to main content
Discord AutoMod can block a scam message and time the account out. It cannot ban. AutoMod Escalation is the missing step: when one of your AutoMod keyword rules catches a message that matches a confirmed scam pattern, Hashbot bans the account, removes everything it posted in the last hour, and posts an evidence card in your log channel. This is a premium feature. The Obfuscated URL scam ban is on by default; enforcing your own rules is opt-in, rule by rule.
Only the confirmed scam pattern is ever banned automatically: an obfuscated link combined with a fake-support lure or obfuscation markers. We reviewed 400 real AutoMod hits by hand before turning this on: 0 false bans.

Out of the box: the Obfuscated URL scam

Every paid server gets the fake-support scam handled without any setup. Hashbot reads each message itself and bans accounts that post the confirmed pattern: a link chopped across lines or disguised with symbols, plus a support or ticket lure. No AutoMod rule and no Manage Server permission needed. Turn it off in Settings, Obfuscated URL Scam, or with /settings obfuscated-url set-state state:disable. Set the AutoMod Escalation punishment to Flag if you want an alert instead of a ban. The rest of this page is about AutoMod Escalation: letting Hashbot also act on hits from your own AutoMod keyword rules.

What it catches

The classic “fake support” copypasta and its variants:
Links chopped across lines, four slashes (:////), ideographic full stops instead of dots, @dis cord splits, percent-encoded hosts (:/#@%64%69%73…, discord:%64…), and junk before the host (https://@, http://:link:@). The one-click Obfuscated URL AutoMod rule (v2, September 2026) also blocks :///, ://@, mailto:/, :/#, nested-quote > > ladders and backslash-chopped invites before they post. Defanged links that members paste when reporting a scam ([.], (dot), hxxp) are not treated as obfuscation. Hashbot scores every keyword hit for these signals plus context (Discord invite, brand-new account, joined in the last day, lure words and emoji). Anything that is suspicious but not the confirmed pattern is only reported if you turn on review alerts.

To escalate your own AutoMod rules you need

1

Give Hashbot the Manage Server permission

Discord only sends AutoMod events to bots that hold Manage Server. Open Server Settings, then Roles, select the Hashbot role, turn on Manage Server and save. The dashboard tells you when it is missing.
2

Have a keyword rule that catches the scam

Any AutoMod keyword rule works. Optionally install the Obfuscated URL AutoMod rule from the AutoMod Rules page: it blocks the message before it posts and times the poster out, on top of Hashbot’s own ban. Then switch on Enforce for that rule.
3

Set a log channel

Ban cards and review alerts go to your Hashbot log channel. Without one, Hashbot falls back to a DM to the server owner.

Switching enforcement on

Hashbot enforces nothing until you switch it on, rule by rule. On the AutoMod Rules page every keyword rule has an Enforce switch. Switch it on for the rules that catch scams (the Obfuscated URL rule, your scam-keyword rules); leave self-promo or profanity rules off. A rule that is disabled in Discord cannot enforce until you enable it there.

Punishment and review alerts

At the top of the AutoMod Rules page, the Escalation section has two controls:
  • Punishment: Ban (default) bans the confirmed pattern. Flag posts it to your log channel with the full message and two buttons, Ban and Dismiss, so a moderator decides in one click.
  • Review alerts: off by default. When on, Hashbot also posts suspicious hits that are not the confirmed pattern (members re-posting an invite, a link with a lure word). About 1 in 6 of those is a real scam, so most servers leave this off.

What happens on a ban

  1. The account is banned with the reason AutoMod Escalation: rule "<name>" matched "<keyword>"; <signals>; score <n> (the out-of-the-box detection has no keyword part).
  2. Its messages from the last hour are deleted.
  3. A card lands in your log channel with the rule, the keyword, the score breakdown, member context and an Open Dashboard button. Flag cards also quote the full message, and a dismissed Flag card is stamped with who dismissed it.
  4. The ban is in your dashboard ban log with a one-click Unban.
Staff, allowlisted members and members above Hashbot’s role are never banned. A server that bans more than 25 members through enforcement in 10 minutes is paused automatically and the owner is notified.

Slash commands

include-rule switches a rule on, exclude-rule switches it off.

Frequently asked

Will it ban my members? Only for the confirmed pattern: an obfuscated link plus a fake-support lure or obfuscation markers. Uncertain hits never ban; with review alerts on they are posted for a moderator to decide. Does it replace AutoMod? No. For the Obfuscated URL scam Hashbot deletes the message itself. For your own rules, AutoMod blocks the message and Hashbot finishes the job. I disagree with a ban. Open the dashboard from the card, find the ban under Logs and click Unban. If you think the pattern is wrong, tell us in the Hashbot server with the banned user’s ID and your server name.