Frostbite
Anti-raid vs anti-scam bots: what Wick and Beemo miss
Raid protection stops a flood of new accounts. Scam detection stops a trusted member posting a fake giveaway. Different problems, and most servers need both.
Two different attacks
A raid is volume. Dozens or thousands of accounts join at once and flood channels, mass-mention, or try to nuke the server. The signal is the join pattern: account age, join rate, shared avatars, and how fast they act after joining.
A scam is trust. One account that has been in the server for months posts a fake Nitro gift or a giveaway screenshot. There is no join spike, no new accounts, no mass mention. The signal is the content of the message and, in the hacked-account case, the same content appearing in several channels within seconds.
A bot tuned for the first attack has no reason to look at the second. That is not a flaw. It is scope.
What the raid specialists do
Beemo watches joins for the signature of a userbot raid and bans the whole batch, with no configuration beyond granting Ban Members. It does that one thing and stays quiet otherwise.
Wick covers anti-nuke, anti-raid, anti-spam, verification, and a broad moderation toolkit behind a dashboard. It looks at member behaviour, role and channel changes, and join floods.
Both are widely used, both are worth running, and neither fingerprints images, reads the text inside a screenshot, or notices that a long-standing member's account just started posting the same picture in five servers.
What a scam specialist does
Frostbite only looks at message content and cross-channel patterns:
- Perceptual fingerprints of known scam images, shared across every server running it, so a resized repost is still caught
- OCR on new images, matched against known scam phrases, so a screenshot nobody has flagged yet is caught by what it says
- Cross-channel and cross-server detection for the hacked-account wave
- A honeypot channel for whatever the content filters miss
- A recovery DM to the member whose account was used, instead of treating them as the attacker
It does not watch joins, ban raiders, verify new members, or protect roles and channels. If you point it at a raid, nothing happens.
Running them together
They do not overlap, so there is nothing to configure between them. Give the raid bot Ban Members and let it own joins. Give Frostbite View Channel, Read Message History, and Manage Messages so it can see and remove content, and run:
/filter quicksetup log_channel:#mod-logUse the same log channel for both if you like. Frostbite collapses a spray across many channels into one alert, so it will not drown out the raid bot's output.
A rough guide to which you need
| If your problem is | You need |
|---|---|
| Waves of new accounts joining and spamming | A raid bot |
| Someone with admin deleting channels | An anti-nuke bot |
| Trusted members posting fake giveaways and gift links | A scam bot |
| The same scam image appearing across your servers | A scam bot with a shared blocklist |
| All of the above | One of each |
Most servers over a few hundred members eventually meet all of them. The commands page shows exactly what Frostbite adds and what it leaves to the others.