shrkbotv3.8.2
TR
RG

You're looking at a preview - nothing you change here is saved.

Invite shrkbot

Scam Image Detection

Reads the text inside posted images and checks it against known scam patterns and previously confirmed scam images. Staff confirm or dismiss every catch, and the bot remembers.

Enabled

Part of Moderation. Shared settings - staff role and alerts - live on the Moderation overview.

Before you turn this on

Deletion is on by default, and detection is automated and imperfect. Screenshots of scams - the kind well-meaning members post as warnings - can be detected and deleted too, so tell your community not to post scam screenshots. Staff are pinged on every action and can mark a decision as wrong with one click in Discord.

Report a scam yourself

Anyone with the staff role can right-click any message with images -> Apps -> shrkbot -> Report as scam. Works whether or not automated scanning catches it.

Words or phrases that mark an image as a scam when they show up in its text - on top of shrkbot's built-in scam rules. Close misspellings count too, so one entry covers its variations. Short, common words trigger easily; longer phrases make more precise rules. Keywords are matched like all other text - case, punctuation and disguised characters ignored, digits kept.

Recommended default: 2

matching keywords before shrkbot acts

1 is aggressive - a single matched word can delete an image. 2 or more keeps one stray word from triggering. Up to as many keywords as you've set.

Flag only leaves the image up and pings staff to review it.

Applies on top of the purge.

Sets the punishment for images a moderator has confirmed as scam on this server, overriding the general punishment above. Leave on "Same as above" to reuse it. Confirmations from other servers don't count.

How image scanning works