10 filters. One toggle per rule. Zero latency.
AutoMod scans every message in real time with 10 independent matchers — you decide per filter: delete only, warn only, both, or off. Each filter has its own bypass channels. Shared immune roles (inherited from Moderation) protect your team with a single config. The hot path stays under 50ms per message.
What AutoMod can do
Profanity (500 words), repeated text (2–100 chars), server invites (own-server whitelist), external links (domain whitelist + block-all toggle), caps (50–100% threshold), emojis (max 1–100), spoilers (max 1–100), mentions (max + @everyone weighting), zalgo (combining mark threshold), anti-spam (2–20 messages / 1–60 s).
Each of the 10 filters has an independent action toggle. Delete-only removes silently. Warn-only writes to mod history and optionally DMs. Delete+Warn does both. Off ignores the filter entirely.
Up to 50 bypass channels per filter. Exclude NSFW channels from the caps filter, make the spam filter laxer in the chat channel, allow links in #links but nowhere else — full granularity.
Up to 50 immune roles — configured in Moderation, automatically valid for AutoMod too. Users with an immune role trigger none of the 10 filters. One config location, both modules benefit.
One switch turns all 10 filters off at once. For events (giveaway spam is fine), maintenance windows, or emergencies — without touching every individual filter config.
marketing.features.automod.cap_6_desc
In detail
Performance on large servers
All 10 matchers are O(1) without I/O — under 50ms hot-path latency per message. On a 100k-member server with 1,000 messages/minute, AutoMod runs with negligible CPU impact.
Notification cooldown
Warn DMs have a built-in cooldown (1 warning per user / 10 seconds per channel). This means a spammer sending 10 messages in 5 seconds produces exactly one DM ping, not ten.
Bots are never filtered
AutoMod ignores bot messages by default — none of your other bots (Dyno, MEE6, Carl) will trigger our filters. No bot-war loops, no accidentally blocked webhook spam.
Frequently asked questions about AutoMod
Can I explicitly whitelist individual users?
User whitelisting works through roles (immune roles in Moderation, up to 50). Create a "Trusted" role and assign it to the relevant users — AutoMod ignores them entirely.
How does profanity matching work?
Literal string match (exact or partial via toggle) with optional case sensitivity. No automatic leet variations (b!tch → bitch) — add all variants to the list, up to 500 entries.
Why don't delete-only actions appear in the Audit-Log?
Delete-only is intentionally silent — no ModAction record, no DM. If you want to see the action, set the filter to "Delete + Warn" (writes a ModAction) or enable the Audit-Log module separately (logs all message deletes).
Is there a live preview for "would this message be flagged"?
Not currently — it is planned for a future version. Test your filter config on a staging server by sending example messages.
More modules
Ready for TeraOne?
Two clicks to set up. No onboarding, no credit card.