Protection with an audit trail

Moderation your team can defend.

Detection, enforcement, cases, and recovery are one system. Rules act the moment they match, every action becomes a case a human can read, and nothing your moderators do is a dead end.

RULE

AutoMod with an escalation ladder

Flood, spam, and content rules act the moment they match, and repeat offenders move up a ladder you define instead of getting the same warning forever.

LINK

Link filtering with a reputation feed

Known-bad hosts are caught against an ingested reputation list, and a false positive can be reported back and allowed without editing a rule by hand.

GATE

Join gate and verification

New members pass a verification step before autoroles are granted, so a raid never inherits the permissions a normal join would.

NUKE

Anti-nuke and panic mode

Destructive bursts — channel and role deletions, mass bans — trip a threshold and can put the server into panic mode while you sort it out.

CASE

Cases with reasons and attribution

Every action becomes a numbered case carrying the rule that matched, the evidence, who acted, and whether the member was told.

BACK

Reversals and raid cleanup

Undo is first-class: /unban and /untimeout, a durable mass-action runner for cleanup after a raid, and message logs to reconstruct what happened.

Moderator commands

Each one checks that the moderator may act, that Guardy holds the permission it needs, and that the target sits below both of you in the role hierarchy — then records a case.

  • /warn
  • /timeout
  • /untimeout
  • /kick
  • /ban
  • /unban
  • /purge
  • /slowmode

What is not here yet

The AI review queue is a later phase and is not part of moderation today. It is documented separately so nobody has to guess which parts of Guardy are running.

See the AI review queue roadmap