Ember's Anti-Nuke system watches Discord's audit log for 20+ destructive action types (bans, kicks, channel and category deletion, role changes, webhook creation, permission overwrite edits, and more), using a per-action threshold instead of one blanket "X actions = nuke" rule. Every threshold is configurable, and a threshold of zero triggers immediately.
Default Thresholds
Action
Default Trigger
Bans
3
Kicks
3
Channel deletes
3
Channel creates
5
Category deletes
2
Channel renames
3
Role creates
5
Role renames
3
Role deletes
3
Server icon changes
2
Server name changes
2
Adjust any threshold from the &antinuke panel. Once an attack is detected, the server enters a temporary lockdown: thresholds collapse to immediate response, dangerous new bots can be removed on sight, and later waves of the same attack stay tied to the existing incident instead of resetting the clock.
Rapid Response Guard
The audit-log system above has one unavoidable limitation: Discord doesn't always attach an executor to an audit log entry the instant it's created. Rapid Response Guard closes that gap by watching raw gateway events directly, before the audit log resolves who's responsible.
Gateway burst3s window
Emergency lockdown~0ms
Audit log resolvesthe attacker
Punishment + restorationself healing
Three major destructive gateway events (or enough weighted activity) inside a 3-second sliding window trigger an early lockdown immediately, before the audit log even knows who did it. This meaningfully shrinks the window an attacker has before Ember can act.
Threat Detection
Coordinated Bot Attack DetectionEXCLUSIVE
Ember tracks attacking bot IDs across the whole server. If Bot A deletes a channel, Bot B deletes another, and Bot C starts editing roles within the same window, they're classified as one coordinated attack instead of three unrelated incidents.
Unverified Bot Join-WavesEXCLUSIVE
Unverified bots that join within a short interval are grouped into a join wave, even if none of them act maliciously yet. If any single bot in that wave later performs a destructive action, Ember can neutralize the entire wave at once.
Delegated Threat LineageEXCLUSIVE
Ember records who added every bot in the server, then correlates later malicious-bot incidents back to that person — with incident history, unique malicious-bot counting, re-attack weighting, escalation thresholds, deduplication windows, and a fresh whitelist check before any punishment.
Invite-Linked Role Escalation GuardEXCLUSIVE
Discord lets servers attach a role to an invite link. Ember reads invite-to-role relationships from Discord's REST API, reverts a dangerous edit instantly, identifies the audit-log executor, and quarantines or bans them depending on configuration.
Anti-Bypass RoleEXCLUSIVE
Ember maintains a special Anti-Bypass role near the top of its own hierarchy. If someone tries to move or bury it to disable Ember's permissions, the role repairs its own position automatically.
Whitelisted Automation DetectionEXCLUSIVE
Being whitelisted doesn't make an actor invisible. A trusted admin or bot behaving at machine speed (rapid bans, mass role movement, permission escalation, link or ping floods) is still flagged.
Bot Trust & Bot Audit Center
The &bottrust system supports granular trust by action category (Ban, Kick, Channel, Role, Webhook, Emoji, Sticker), so a bot can be trusted for one kind of automation without being trusted for everything. &bots opens the Bot Audit Center: every bot in the server, classified by verified status, trust level, dangerous permissions, and join info, with filters for dangerous or untrusted bots.
Security Center
Config Audit
Flags conditions like Anti-Nuke disabled, quarantine not configured, dangerous roles above Ember, or unpunishable bot roles: a posture audit, not just a settings list.
Attack SimulationEXCLUSIVE
A non-destructive diagnostic that checks role hierarchy, quarantine viability, and whether external bots are bannable, without actually nuking anything.
Risk Report
Scores overall security posture from weaknesses like missing logs, weak hierarchy, or disabled modules, and explains each deduction.
All three live under &security, alongside a moderator-command anti-spam guard: staff can burst moderation commands during an active raid, but repeated bursts outside that context trigger a progressive lockout (3, then 10, then 30 minutes), exempting the owner and anyone already whitelisted.
Granular Action Whitelists
Grant a trusted admin or bot a bypass for one action type instead of every permission Anti-Nuke controls:
all
Full anti-nuke bypass
ban
Bypass ban protection
kick
Bypass kick protection
channeldelete
Delete text/voice channels
channelcreate
Create new channels
channelupdate
Modify channel settings
roledelete
Delete server roles
rolecreate
Create server roles
roleupdate
Modify role permissions
memberroleupdate
Assign staff roles
guildupdate
Change server settings
botadd
Invite external bots
webhook
Manage webhooks
Command Reference
Command
Syntax
Description
Permissions
Example
&antinuke
&antinuke or /antinuke
Opens the interactive Anti-Nuke Control Panel to toggle real-time protection, manage per-action thresholds, and configure settings access.
Server Owner or Settings Whitelisted Admin
&antinuke
&whitelist
&whitelist <@user/@role> or /whitelist
Opens the Whitelist Management panel to grant or revoke Anti-Nuke bypasses, either globally or for one specific action type.
Server Owner or Settings Whitelisted Admin
&whitelist @Moderator
&bots
&bots or /bots
Opens the Bot Audit Center: lists every bot in the server with verified/unverified status, trust level, dangerous permissions, and join info.
Administrator
&bots
&bottrust
&bottrust <@bot> <action> or /bottrust
Grants a bot trust for specific action categories (Ban, Kick, Channel, Role, Webhook, Emoji, Sticker) without trusting it for everything.
Server Owner or Settings Whitelisted Admin
&bottrust @MyUtilityBot channel
&incident
&incident [list|view <id>] or /incident
Opens the Incident Center: browse, filter by subsystem or severity, and view the full evidence trail behind any security action Ember has taken.
Administrator
&incident list
&security
&security or /security
Opens the unified Security Center: current threat state, lockdown status, AutoMod and Anti-Nuke status, and a security configuration audit.
Administrator
&security
&commandaccess
&commandaccess
Opens the Granular Command Access panel to grant staff members access to specific command groups without full Administrator permissions.