Antinuke
Stop nukes and hacked accounts, whitelist, extra owners and safety alerts.
Antinuke watches the audit log and stops people - or hacked bots - who delete channels, roles, ban members, add bots, create webhooks and so on. It's free for every server.
Quick start#
- Put Miwi's role at the very top (Server Settings → Roles) and give it Administrator. Miwi can only act on members below it.
/antinuke enable- a four-step setup: choose what to protect, the punishment, a log channel (or let Miwi create one) and who to whitelist, then test mode (Miwi logs what it would do) or protect now.- Whitelist more later:
/antinuke whitelist add @YourModBot(?wl add @bot). /antinuke settingsshows every protection with an On/Off button beside it - tap one to switch it. After a day of reviewing/antinuke logs, go live:/antinuke test-mode enabled:false.- Optional:
/owner extra add user:@co-ownerfor someone who helps run security.
Guard role#
Someone with Administrator who isn't whitelisted could try to re-add Miwi through its invite with Administrator unchecked - that rewrites Miwi's own role and used to take its power away.
So setup gives Miwi a second role, Miwi Guard, placed right under Miwi's own role. It has Administrator (or, if Miwi doesn't have Administrator, every permission Miwi has), so Miwi stays armed whatever happens to its own role.
- Anyone who isn't trusted and takes the guard off Miwi, removes its permissions, deletes it or weakens Miwi's own role is punished at once.
- Miwi then puts the guard back - re-added, permissions restored, or recreated.
- The owner, extra owners and whitelisted members can still change it.
- For the strongest guard, give Miwi Administrator and run
/antinuke enableagain.
How fast it reacts#
- Straight from the audit-log stream. Each action arrives with who did it, so there's no polling and no "latest entry wins" guessing. Delayed entries are correlated by type, target and time.
- Containment and clean-up run in parallel. The attacker is punished while their channels, roles, webhooks and bots are removed or restored. Floods (created channels, roles, webhooks, bots) are deleted first.
- Critical changes act on the first occurrence, whatever the threshold: vanity URL changes, 2FA-requirement changes, and dangerous permissions (Administrator, Manage Server…) given to @everyone.
- Strict defaults: one dangerous role grant, one bot added, one prune or one integration by an untrusted member is enough. Change any threshold with
/antinuke protection. - Administrator is not trust. Only the owner, extra owners and whitelist entries are trusted.
The trust model#
| Tier | Who | What it means |
|---|---|---|
| 👑 Owner | The server owner | Always trusted. Manages everything. |
| 🔑 Extra owner | Added by the owner (/owner extra add) | Whitelisted for every action, can manage security and the whitelist, opens the dashboard, can't be moderated by anyone but the owner, and gets safety alerts. Max 10. |
| ✅ Trusted | Users, roles or bots on the whitelist | Not punished for actions in their scope. Optionally protected (regular mods can't moderate them) and AutoMod bypass (on by default). Can be temporary. |
Administrator is not trust. A hacked admin account is exactly what antinuke is for.
Commands#
Everyday commands (/antinuke)
| Command | What it does |
|---|---|
/antinuke enable | Guided setup (safe to re-run to change anything) |
/antinuke settings | Every protection with its on/off switch, punishment, log channel, whitelist size |
/antinuke disable | Turn antinuke off |
/antinuke whitelist add user: · remove user: · list | People and bots antinuke never punishes |
Advanced (/antinuke)
| Command | What it does |
|---|---|
/antinuke settings | The overview plus the last 7 days and lockdown state |
/antinuke enable · /antinuke disable · /antinuke test-mode enabled: | Switch it on/off; test mode on/off |
/antinuke protection group: … | Per group: on/off, threshold, window, punishment, revert |
/antinuke whitelist add target: scope: level: duration: protected: bypass: note: | Whitelist a user, role or bot |
/antinuke whitelist remove target: · /antinuke whitelist list | Manage the whitelist |
/antinuke logs | Recent decisions with reasons |
/antinuke lockdown start reason: · end · status | Emergency: lock every channel for @everyone |
/antinuke snapshot create · list · diff · restore | Server structure backups (channels, roles, permissions) |
Scopes: Everything · Channels & overwrites · Roles & dangerous permissions · Bans, kicks & prunes · Bot additions · Webhooks · Emojis & stickers · Server settings, integrations, events, AutoMod · Bulk message deletion. Adding a second scope to the same entry widens it.
Prefix aliases: ?an (= antinuke), ?wl add @user, ?extraowner, ?eo.
Safety alerts#
When antinuke acts (or would act in test mode), a raid starts, or a lockdown begins, Miwi DMs the owner and every extra owner - with a link to the log and a Mute alerts for this server button. Alerts are throttled so an attack doesn't flood your DMs.
/owner alerts server:false- off for everyone (owner only)/owner alerts me:false- mute just yourself/owner alerts test:true- send a test alert
Dashboard#
Security page: protections, the whitelist (scope, expiry, protection, AutoMod bypass), recent events. Only the owner and extra owners can change it. Access & alerts page: the alert switch.
Troubleshooting#
| Problem | Fix |
|---|---|
| "Antinuke could not fully stop an attack" | Miwi's role isn't high enough or lacks Administrator. Run /owner scan. |
| A trusted bot got punished | /antinuke whitelist add trusts it for everything; a scoped /antinuke whitelist add entry only covers its scope. |
| Nothing happens | Still in test mode, or the protection for that group is off: /antinuke settings. |
| Unattributed events | Discord didn't write an audit log entry; Miwi never punishes without one. |
Still stuck? Ask in the support server - real people answer.