Layer 1: per-person anti-flood (every plan)
Watches each user separately, by account, IP and device. Anyone writing too fast or repeating the same thing gets muted briefly, and repeat offenders get longer mutes. Your moderation team is never affected.
Layer 2: automatic anti-raid (Pro and above)
Looks at the chat as a whole and estimates the risk from signals such as how many different people are breaking the rules at once or how many messages are exact copies. A crowd arriving on its own never triggers anything.
| Level | What changes |
|---|---|
| 🟢 Normal | Everything works as usual. |
| 🟡 Watch | Slightly stricter limits for people writing very fast. |
| 🟠 Alert | Strict limits and automatic guest-to-member promotions are frozen. |
| 🔴 Anti-raid | Guests become read-only. Registered users keep talking as usual. |
Once the attack is over, protection drops one level at a time after a calm period. If you see an attack coming, you can force the top level manually for 15 minutes. Every incident is logged with its duration and the number of identities sanctioned. Exact thresholds are not published so they’re harder to game.
Frequently asked questions
Can anti-raid block my real audience?
A crowd on its own doesn’t trigger it. At the top level only guests become read-only; registered users and your team keep writing.
Can I turn it off?
Automatic protection is always on, but you can force the top level manually for 15 minutes.