Moderation

By default, links are only allowed from domains on your server's whitelist. A link to any other domain is removed and the member is warned. Add the domains you trust (your website, socials, docs) to the whitelist in your moderation settings so they're never touched.

Each removed link adds a warning. When a member reaches the warning threshold — 5 by default, configurable per server — they're automatically banned.

Suspicious links are scored for risk, and the action scales with the score:

• High risk (score 60+) — deleted immediately, the member is timed out, and a scam alert is raised on your Security Alerts dashboard.
• Medium risk (score 35–59) — deleted and the member is warned (no timeout); the same warning threshold applies, so repeated medium-risk links lead to a ban.
• Lower risk — allowed, subject to the link-whitelist rule above.

When the built-in scoring is uncertain, an AI classifier is used as a fallback signal to help judge whether a link is malicious.

If a member has 3 or more high-risk URLs deleted within 24 hours, Power Horse raises a repeat-offender alert. They're surfaced on the Security Alerts dashboard where you can Ban, Kick, or Mute them in one click.

Every new member is scored on join using signals like account age, whether they use a default avatar, and whether the username looks auto-generated (e.g. lots of trailing digits). The risk level is logged to your audit log and shown in your mod-log when they join, so you can spot likely bot or raid accounts early.

Members can report suspicious links via the report-scam channel; reports get an AI-assisted severity rating and appear in your dashboard for review. You can also opt in to the cross-server threat pool — a shared list of flagged domains so that a scam caught in one community helps protect the others.