Anti-Nuke

Raidschutz (raid protection) watches who comes in. This one watches what someone does who is already inside and has permissions.

This is the case nobody has on their radar and the one that finishes off a server in two minutes: a moderator whose account has been taken over, or one who leaves and tidies up on the way out.

How it works

Blunt and counting: how often has this person carried out this kind of intervention within this time window? No guessing at intentions — a number, a limit, a consequence. Anything else would not be explainable, and protection nobody can explain gets switched off.

The interventions that really hurt are enabled by default:

Intervention	Limit	Time window
Deleting channels	3	60 seconds
Deleting roles	3	60 seconds
Banning members	5	60 seconds
Pruning members	1	5 minutes
Changing role permissions	5	60 seconds

With Alles (everything), creating channels and roles, kicking, webhooks, bots and emojis are added as well. Twelve interventions in total, each with its own limit.

What happens then

Quarantäne (quarantine) is the default, and for a reason: it stops things immediately and can be undone in ten seconds. All roles are removed and remembered:

/schutz freigeben mitglied:@Name

gives them back. A ban would not be so easy to reverse — and on the first false alarm that is exactly the difference between "annoying" and "one member fewer".

If you would rather let it run alongside first, set it to Nur melden (report only) until the limits fit. Kicking and banning are available too, but they only make sense once the protection has proven itself.

The threat level

One switch that tightens all limits at once:

/schutz stufe stufe:erhöht minuten:120

Level	What it does
Normal	The configured limits apply
Erhöht (raised)	All limits halved, medium interventions are monitored as well
Abriegelung (lockdown)	Every monitored action triggers on the second occurrence

The use of this shows itself on the evening when something is going on but it is not yet clear what: instead of lowering twelve limits one by one, you set the level one step higher.

With minuten: it drops back on its own. Use that — a permanently raised level eventually only gets in the way of your own people.

Trust

Anyone entered as trusted never triggers it. Enter yourself before your next big rebuild — otherwise you will lock yourself out of your own server while tidying up.

With trusted roles, caution is advisable: that same role is held by whoever's account gets taken over.

What the protection cannot do

Stop the server owner. Discord does not allow a bot to act against the owner. In that case the protection reports it, and that is all — and it says so plainly.

People above the bot role. Drag the Fishbot role further up in the role list, otherwise the protection will not take effect for exactly the people it ought to.

Without the "Audit-Log einsehen" (view audit log) permission it comes to nothing. The audit log is the only source that also tells you who did something — the "channel deleted" event alone does not. /schutz stand gives a clear warning if the permission is missing.

After a restart

The counters start over. That is the right behaviour: an attack that stretches across a restart triggers again on the second wave — and a half-forgotten counter would otherwise hit someone uninvolved.

Commands

Command	What it does
/schutz stand	What currently applies, including a warning about missing permissions
/schutz stufe	Set the threat level, optionally with a time limit
/schutz freigeben	Take someone out of quarantine
/schutz liste	Who is currently in quarantine
/schutz setup	Limits, action, reporting channel, trust