-
Notifications
You must be signed in to change notification settings - Fork 0
How protection works
The AntiLink scanning pipeline - what gets checked, who is exempt, and how prohibited links are handled.
View this page in the localized AntiLink documentation
Understanding how AntiLink processes a message makes setup, whitelisting, and troubleshooting far easier. Here is what happens to every message.
For each message posted in your server, AntiLink runs these steps in order:
If the author is the server owner, an administrator, a whitelisted user or role, or the message is in a whitelisted / ignored channel or category, AntiLink stops here and does nothing. Exempt authors are never moderated.
AntiLink looks for URLs and Discord invites in the message text. A message with no links is ignored (file uploads and plain text are not links).
The link is classified as a Discord invite, an external link, or a known-malicious / phishing URL, and checked against your enabled protections and your whitelist. A whitelisted domain is always allowed.
The link is risk-scored, then the message is handled according to your enforcement mode (delete, quarantine, or warn), and a moderation log entry is recorded.
This is the single most important thing to understand, and the most common source of "it isn't working" reports:
Warning
AntiLink always exempts the server owner, anyone with Administrator, Manage Server, or Manage Messages, your whitelisted users and roles, and whitelisted or ignored channels and categories. If you test with your own admin account, nothing will happen - that is by design. Test with a normal member account.
This safety-first exemption applies across every module: link protection, Automod, Scam Guard, and Honeypot channels.
| Signal | Used for |
|---|---|
| Message content | Finding links to moderate |
| Message author + roles | Checking exemptions and whitelist |
| Channel / category | Checking whitelisted or ignored locations |
| Join timestamp (from the message) | New-member link lockdown |
AntiLink does not store your messages. When a link is removed, it logs the author, channel, reason, and the detected link for your dashboard activity view. See Security & Trust for the full data breakdown.
When a prohibited link is found, the enforcement mode decides what happens
(configure it with /config or the dashboard Protection tab):
| Mode | What it does |
|---|---|
| Delete | Removes the offending message and logs it. The default. |
| Quarantine | Removes the message and applies your mute role, holding the user for review. |
| Warn | Logs a warning and escalates repeat offenders. |
In Warn mode, repeat offenders climb a ladder rather than being punished all at once: a warning first, then the mute role, then progressively longer timeouts. Counts reset after 30 days of good behavior.
The /checklink command and the dashboard rate a URL as
clean, low, medium, high, or critical, using offline
heuristics plus optional threat-intelligence feeds. "Clean" means no known threat
was found at the time of the check - it is never a guarantee of safety.
Info
Ready to configure it? Follow the Setup Guide, then see Best Practices for recommended roles and whitelist strategy.
Website | Dashboard | Documentation | Status | Support