Skip to content
Axaro

Security

Live

Raid mode, message waves, anti-nuke and impersonation.

How a raid is spotted

Axaro computes a score over the joins of a sliding window. An ordinary join counts for 1; each suspicious signal adds 1, up to 4 per account. When the score passes the threshold, raid mode opens.

SignalWhat it spots
recent accountCreated less than 24 hours ago (adjustable)
same batchCreated within minutes of at least two other joiners
serial nameSame name start as at least two others: raidbot_83jd2, raidbot_k2m1s

Three real raid accounts are enough to pass the default threshold, while a genuine wave — a video that takes off, a partnership — brings older, varied accounts and stays below it.

Raid mode

A raid is not a moment, it is a period. During raid mode the server is closed and every account that joins is handled, including those arriving after the trigger.

  • Invites paused — nobody can get in. It is the only lever that stops the joins themselves: verification stops people from writing, not from entering. Discord limits the pause to 24 hours.
  • Verification at the highest level — only accounts with a verified phone can write.
  • DMs between members paused, if you turn it on — against raids that target members one by one in private.

The action on accounts is yours to choose: nothing, timeout, kick or ban. It targets suspicious accounts, or every new account.

Raid mode stretches while joins keep coming, then everything reopens. Axaro only gives back what it closed: if your team changed the verification or extended the pause by hand during the raid, their choice is kept.

/security lock opens raid mode by hand, /security unlock ends it and gives the summary. /security status says what is closed, and until when.

Clean up in one click

The alert posted in the alert channel carries three buttons, for the team that gets it on their phone in the middle of the night.

ButtonEffect
Ban the raid accountsBans the whole wave, two hundred accounts per call, and deletes their messages from the last hour
End raid modeReopens the server right away
False positiveEnds raid mode, lifts the automatic timeouts and bans
The buttons only go to the alert channel, not to the owner’s direct message: set an alert channel. The permissions of whoever clicks are checked on every click.

Message waves

Another kind of raid: accounts already on the server, sometimes for days, all posting the same link or the same ping within seconds. AutoMod does not see it — each of them posted only once.

Axaro recognises the same link even when the text around it changes, and even with different parameters. Once the wave is spotted, the messages are deleted, the authors timed out if you choose so, and the same content is blocked for a few minutes.

Staff and members who have been here for a month are never counted. Neither are GIFs or links to a Discord message: people share those together without it being a raid.

Anti-nuke

A compromised moderator account does not delete one channel: it deletes forty in ten seconds. Speed is what separates destruction from a cleanup, so that is what Axaro watches — not the nature of the action.

Action watchedWeight
Channel or role deletion2
Ban2
Webhook creation2
Server settings change2
Channel or role creation, kick1

A deletion weighs more than a creation: creating forty channels is annoying, deleting forty is irreversible. The weight avoids having two thresholds to express that difference.

The source is the audit log, the only place that names who did an action. Without the View Audit Log permission, the protection cannot see anything.

What Axaro does to the culprit

Past the threshold, Axaro removes from the actor the roles carrying dangerous permissions — administrator, managing the server, channels, roles, webhooks, banning, kicking — then records the incident and alerts the team.

Removing roles rather than banning is deliberate: a compromised moderator account belongs to someone on the team. Roles can be given back; a ban has to be negotiated.

  • The server owner and Axaro itself are never targeted — sanctioning the former would be as impossible as it is absurd.
  • A trusted list exempts the team’s legitimate bots: without it, the anti-nuke would disarm their tool instead of protecting the server.
  • Removing permissions automatically is the only reaction that truly protects: without it, the anti-nuke only narrates the destruction while it happens.

Available settings

Anti-raidDefault
joinThresholdScore of 10
windowSeconds30 seconds
actionClose the server, without touching accounts
targetSuspicious accounts
lockdownMinutes15 minutes, 0 to end it by hand
youngAccountHours24 hours
pauseInvitesOn
raiseVerificationOn
pauseDmsOff
Message wavesDefault
minAccounts5 accounts
windowSeconds20 seconds
actionDelete and time out
Anti-nukeDefault
enabledOff — turn it on once the trusted list is filled
actionThreshold5 points over the window
windowSeconds60 seconds
stripPermissionsOn
Discord anti-raid and anti-nuke bot · Axaro