How to protect your Discord server from nukes and raids

A nuke is when someone with too much power — a hacked staff account, a malicious bot or a moderator who turned on you — deletes channels and roles, bans members and spams webhooks within a minute. A raid is the opposite direction: dozens or hundreds of fresh accounts join at once to flood your channels. Both are over before a human can react, so protection has to be in place beforehand. This guide covers the Discord settings that prevent most nukes, the habits that keep you safe, and how BotGG’s anti-nuke and anti-raid respond automatically.

How servers actually get nuked

Almost every nuke starts with one of these:

• A staff account gets hacked — a fake gift link, a token-stealing download or a “try my game” scam. The attacker logs in with the account’s permissions.
• A bot with Administrator gets compromised or was malicious from the start. A bot’s token is a full login.
• Someone was trusted too quickly — a new moderator with Administrator or Manage Roles who decides to wreck the server.
• Permissions crept up — over the years, roles collected Ban Members, Manage Channels and Manage Webhooks “just for this one thing”.

The fix is the same for all four: fewer people with dangerous permissions, a second factor on every staff account, and something that notices mass actions faster than you do.

Step 1: Audit dangerous permissions

Open Server Settings → Roles and go through every role. These permissions can destroy a server and should be held by as few roles as possible:

Practical rules:

• Administrator only for the owner and one or two co-owners. Moderators don’t need it.
• Moderators get Kick, Ban, Timeout and Manage Messages — not Manage Roles, Manage Channels or Manage Webhooks unless their job really requires it.
• Bots get only what they use. If a bot asks for Administrator “to make setup easier”, give it the specific permissions instead. BotGG, for example, doesn’t ask for Administrator.
• Remove roles from people who are no longer active staff. Inactive accounts are easy targets because nobody notices when they’re hijacked.

Step 2: Require 2FA for moderation

In Server Settings → Safety Setup, turn on Require 2FA for moderator actions (only the owner can do this, and the owner needs 2FA on their own account first). From then on, members without two-factor authentication can’t ban, kick, delete messages or manage channels, even if their roles allow it.

It doesn’t stop a stolen session token, but it blocks the most common case: a reused or phished password.

Step 3: Put your protection bot’s role in the right place

A bot can only act on members whose highest role is below its own. If an attacker’s role sits above your protection bot, the bot can see the attack but can’t stop it.

• Drag BotGG’s role above all staff roles that hold dangerous permissions (Moderator, Admin), keeping only the owner’s own role and maybe a co-owner role above it.
• Never give regular staff a role above the protection bot.
• Check again after adding new roles — new roles are created at the bottom, but someone may move them.

Step 4: Watch the audit log

Discord’s audit log (Server Settings → Audit Log) records who deleted, created, banned or changed what. Make a habit of reading it after every staff change, and turn on a log channel so you don’t have to: BotGG Logs posts channel, role, member and moderation events to a private channel in real time.

BotGG’s anti-nuke reads the audit log too — that is how it knows who deleted a channel, not just that a channel was deleted. Without the View Audit Log permission it falls back to a weaker guild-wide mode that can only alert, and the console tells you to grant the permission.

Step 5: Turn on anti-nuke

Open https://console.botgg.app, choose your server, then Moderation → Protection → Anti-nuke. It counts actions per person within a time window (30 seconds by default) and steps in when someone crosses a limit:

Every limit can be adjusted or switched off. Three triggers fire on the first occurrence by someone who isn’t trusted: a role gaining (or a member being given) a dangerous permission like Administrator, Manage Roles or Ban; a bot being added; and the server name or vanity URL being changed.

What happens to the attacker is your choice:

• Quarantine (default) — every role BotGG can remove is taken away and a Quarantine role with no access is added. The removed roles are saved, so you can give them back with one Restore roles click if it was a false alarm.
• Strip roles — remove their roles, optionally with a 24-hour timeout.
• Kick or Ban.
• Alert only — tell staff and do nothing else.

A bot that triggers anti-nuke is kicked instead, since its power is its own managed role.

With revert on, BotGG also undoes what it can: deletes webhooks the attacker created, takes back dangerous permissions and roles they handed out, kicks bots they added and puts the server name back. Each alert says what was reverted and what couldn’t be (for example, deleting webhooks needs Manage Webhooks, which isn’t in BotGG’s default invite).

Trusted users and roles are never acted on. The owner and BotGG are always trusted; add co-owners or a senior-admin role yourself. A newly given trusted role only counts after it has been held for 10 minutes, and an untrusted member handing out a trusted role triggers anti-nuke immediately — so an attacker can’t make themselves “trusted” on the fly.

Kicks and bans that staff make through BotGG (/ban, /kick, the console) are counted against the staff member who asked for them, so a hijacked moderator account can’t get around the limits by using the bot.

Step 6: Turn on anti-raid and the join gate

On the same Protection page, Anti-raid reacts to join floods:

• Detection — N joins within T seconds (10 in 10 by default), plus smart detection that scores bursts by young accounts, default avatars, similar usernames and joins through the same invite.
• Lockdown — @everyone loses Send Messages in the chosen channels (or every public text channel); the previous permissions are restored exactly when the raid ends.
• Newcomers during the raid — if Verification is on, everyone must solve its captcha while the raid lasts; otherwise they get a timeout or a kick. Accounts younger than a set age are kicked.
• The raid accounts — listed in an alert with Ban all listed, Kick all and Dismiss buttons, or kicked or banned automatically if you choose so.
• Auto-end after a number of calm minutes (15 by default).

The join gate works on every join, not only during raids: kick or time out accounts younger than N days or without an avatar.

For a full walkthrough of verification and raid settings, see the verification guide.

Step 7: Keep backups you can restore

Anti-nuke stops an attack within seconds, but a few channels may already be gone by then. Backups snapshot your roles, channels, categories, permission overwrites and server settings, and can restore them later with a dry-run preview first. Restoring needs the owner or an Administrator.

• The free plan stores 1 backup; Pro stores 20 and adds scheduled daily or weekly backups.
• When anti-nuke is first enabled, BotGG takes a safety snapshot, and while Backups is enabled it keeps a rolling daily one.
• Anti-nuke auto-restore recreates deleted channels and roles from the newest backup right after the attack, including the permission overwrites that referenced a deleted role, and posts a report.

Take a manual backup with /backup create whenever you finish a big change to the server’s structure.

Step 8: Have a panic button

If something is clearly going wrong and you don’t know who is behind it yet, /panic start locks every writable channel at once, switches newcomers to captcha when Verification is on, and can quarantine a specific member. If BotGG has Manage Server (the default invite link doesn’t grant it), it also pauses invites for 24 hours. /panic end restores everything. The Protection overview in the console has the same button, plus a protection score that lists what is still missing in your setup.

Recommended settings

FAQ

What is the difference between a raid and a nuke?

A raid comes from outside: many new accounts join and spam. A nuke comes from inside: an account or bot that already has permissions destroys the server. Anti-raid handles the first, anti-nuke the second.

Can anti-nuke stop the server owner?

No. The owner can’t be restricted by any bot — Discord doesn’t allow it. Protect the owner account with 2FA and never share it.

Will anti-nuke get in the way of normal moderation?

The default limits are well above normal moderation (for example five bans in 30 seconds). If your team bans spam waves by hand, add your senior staff role to the trusted list or raise the ban limit.

Is anti-nuke free?

Yes. Anti-nuke (including auto-restore from backups), anti-raid, the join gate and the panic button are part of the free plan. Only scheduled backups and a larger backup history need Pro.