A match is not a complete judgment
Quoted abuse, reclaimed terms, education, and evasive phrasing can require human interpretation. Test edge cases and make appeals possible.
DISCORD AUTOMOD CHECKLIST
Native filters are most useful when triggers, actions, exemptions, and alert ownership are explicit. Use this checklist to review the configuration—and keep human moderation responsible for context.
10 RULE CHECKS
A rule is not finished when a keyword is entered. Its enabled state, actions, exemptions, alert destination, language coverage, and owner determine whether it behaves as intended.
Assign a named owner and backup who understand the community policy. Discord currently requires Manage Server or Administrator to edit AutoMod, and the editing experience is available on desktop.
Record each rule name, trigger type, enabled or disabled state, actions, custom response, exempt roles, exempt channels, and alert destination. Keep intentionally disabled rules visible in the review.
List the abuse patterns AutoMod should interrupt: harmful terms, community-specific phrases, obvious spam, and disruptive mass mentions. Separate firm policy from words that need context.
Choose the preset categories that fit the server, then examine exemptions and likely false positives. Treat Discord's lists as a starting point, not a complete policy for every language or community.
Add server-specific words and phrases, known evasions, and local-language terms. Keep related terms grouped under clearly named rules so moderators can understand why a message was blocked.
Enable the protections that fit the participation model and choose a mention threshold that catches disruption without blocking legitimate staff or event workflows. Test ordinary messages as well as obvious abuse.
Decide when to block the message, send an alert, or apply a timeout on a supported rule type. A block prevents one matching message; an alert creates review work; a timeout affects the member and deserves a documented policy.
Send AutoMod alerts to a private, actively watched channel. Limit access to the people who need moderation evidence and make sure the channel is not exposed by an unexpected role or overwrite.
Require a reason for each exemption and test its reach. A channel exemption also covers threads and text chat in voice channels associated with that channel, so broad exceptions can create blind spots.
Use harmless test strings in a controlled channel, confirm the expected action and alert, record known limitations, and review rules after policy changes, new bots, language shifts, or repeated false positives.
CHOOSE THE RIGHT CONTROL
Layer narrow controls instead of forcing one rule to do everything. The table reflects Discord's current native options; confirm the product interface before relying on a detail in an operating policy.
| Rule family | Useful for | Review closely |
|---|---|---|
| Commonly Flagged Words | Discord-maintained preset categories for insults and slurs, sexual content, and severe profanity | Policy fit, false positives, exemptions, and the current English-language limitation |
| Custom Keywords | Community-specific terms, phrases, and custom words in any language | Context, variants, maintenance ownership, actions, and rule naming |
| Block Spam Content | Messages Discord's filter identifies as spam | Missed spam and false positives; Discord says the filter may not catch everything |
| Block Mention Spam | Messages containing more mentions than the configured limit | A threshold that fits legitimate announcements, events, and staff workflows |
Blocking a message gives the sender immediate friction. Sending an alert creates evidence for moderators, but only if the destination is private and watched. Discord also supports timeout as an AutoMod action for compatible keyword or mention-spam trigger types; the acting bot or user needs the Moderate Members permission.
Write down what a moderator should do after an alert: ignore a confirmed false positive, contact the member, preserve evidence, escalate, or adjust the rule. Without that decision, an alert channel becomes an unread queue instead of a control.
An exemption is not a footnote. Discord allows exempt roles and channels, and channel exemptions extend to their threads and text chat in associated voice channels. Review each exception as access: who receives it, where it applies, why it exists, and when it should expire.
Discord also documents platform-level gaps. Members with Administrator or Manage Server are exempt from AutoMod filters, and messages from apps installed to the server are not flagged. Review those accounts, roles, and apps separately instead of assuming AutoMod covers them.
THE HONEST BOUNDARY
Discord explicitly notes that spam filtering can miss messages. Preset word lists and the spam-content filter are currently English, while custom keyword rules can use any language.
Quoted abuse, reclaimed terms, education, and evasive phrasing can require human interpretation. Test edge cases and make appeals possible.
DisplayState reads supported AutoMod configuration. It does not inspect conversations, observe moderator response, or detect a raid happening now.
Use AutoMod alongside role hygiene, account protection, clear policy, trained moderators, incident procedures, and regular configuration review.
CONTINUE THE REVIEW
A strong rule can still alert the wrong people, exempt an overpowered role, or leave an installed app outside the expected coverage. Review the surrounding access model before calling the work complete.
Use the Discord server security checklist to review verification, @everyone, private rooms, and category exceptions.
Use the Discord bot permissions guide to compare each integration's job with its actual authority and hierarchy.
FREE DURING BETA
Run the free private audit for supported settings, review every limitation, and decide what belongs in your server policy.
Run the free audit