Automated Threat Detection

Three independent, fully automated pattern-detection systems that catch and act on real attacks in seconds - no moderator has to be online or watching for any of these to work.

This is one of Pixie's biggest differentiators from most moderation bots: these three systems run on every message, all the time, and act on their own the instant a real attack pattern shows up - timing out the account and alerting staff before a human moderator could ever react manually.

🚨 Anti MrBeast Scam

$antimrbeastscam

Also $ambs. First run walks through setup; after that it reopens as a live settings panel with four controls: on/off, alert channel, auto-delete the offending images, and a list of roles to exclude from detection entirely (useful for a bot/giveaway-announcer role that legitimately posts images in bursts).

What it catches: the well-known "MrBeast giveaway" scam pattern - a compromised or scripted account rapidly cross-posting the same bait image into multiple channels. When it fires, the account is timed out, staff get an alert naming the member and channels hit, and (if enabled) the offending messages are deleted automatically.

A real alert, with one-click Untimeout/Kick buttons for staff right there.

⛔ Anti Seller

$antiseller

Also $aseller, $sellerscan. First run walks through setup; after that opens a settings panel (on/off, alert channel, excluded roles).

What it catches: off-platform advertising by predators soliciting content from minors - a serious, real-world abuse pattern, not a spam nuisance. Messages are analyzed against a tuned set of language signals rather than a simple keyword blocklist, so it catches coded/evasive phrasing a flat word list would miss. Suspicious messages get deleted and alert staff; the more serious the pattern, the more automatic the response.

A real alert, with one-click Untimeout/Kick/Ban buttons for staff right there.

🛡️ Anti Generic Hacked Account Spam

$antighspam

Also $aghs. First run walks through setup (pick an alert channel, confirm permissions); after that, re-running the command reopens a live settings panel - toggle it on/off, change the alert channel, toggle auto-deleting offending messages, and add/remove excluded roles, all from the same buttons, same as Anti MrBeast Scam's and Anti Seller's panels.

What it catches: the classic sign of a hacked/stolen account - the same scam-link message getting blasted into multiple channels faster than a person would type it. When it fires, the account is timed out and (if enabled) every copy of the spam message is deleted automatically.

A real alert - both copies were auto-deleted and the account timed out on the spot.