Automod triggers
Every trigger you can add to an automod rule, what it looks at and its default limits.
Words, links, invites and files
Block keywords matches words or phrases you list. Block regex patterns matches regular expressions. Block or allow URLs takes an allow list and a block list of domains. Block invites blocks every invite link unless you list allowed invite codes or server IDs, and it can allow your own server's invites. Block file types takes an allow list and a block list of extensions, such as .exe and .bat to block or .png and .pdf to allow. Phishing detection has its own guide.
Burst triggers
Message, link, mention, attachment, emoji, line, character and GIF spam each count one kind of content from one member inside a time window. They share a count and an interval, defaulting to 5 within 10 seconds. Line spam can also fire on a single message with more than a set number of newlines. Duplicate message spam counts identical messages and defaults to 3 within 60 seconds. Once one of these fires for a member, Nighthawk only deletes their messages for the next 10 seconds.
Single-message triggers
All-caps message fires when a message of at least 15 letters is mostly uppercase (the default is 70 percent). Zalgo fires when combining marks make up more than a set share of the message, 30 percent by default. Repeated characters fires on a long run of the same character. High mention ratio fires when mentions are more than a set share of the words, 50 percent by default. Mention spam counts mentions over time, while mention ratio judges one message. Excessive newlines still works in older rules but is no longer offered, because line spam covers it.
Member triggers
New member (account age) fires whenever someone joins. Give it a minimum account age such as 3d and it only fires for accounts younger than that. Machine Learning Anti-Raid scores every join and fires when several accounts arrive together and look coordinated. A single member joining alone can never trigger it. The Anti-raid guide covers its settings.
Staff and trap triggers
Moderation action spam (Nuke) counts one moderator's actions and has its own guide, Nuke protection. Honeypot channel fires on any message in a trap channel and has its own guide, Honeypot channels. AI Moderation is managed from the AI Moderation page and cannot be added by hand.
Combine triggers
Add more than one trigger to a rule and it acts when any of them matches. Conditions can apply to every trigger in the rule or only to the ones you pick, so one rule can mix a keyword filter for messages with an account age check for joins.