LavaBot LavaBot Docs
← Main site

Insights & anti-impersonation

Server Insights · a weekly report every Monday:

  • !insights dm · DM it to the server owner.
  • !insights #channel · post it in a channel.
  • !insights now to preview · !insights off to stop.

Anti-impersonation (Ultimate) · flags members whose name or avatar mimics your staff, owner or protected users. The engine sees through look-alike tricks: cross-alphabet homoglyphs, leetspeak (P0lloSpe1lat0 vs PolloSpellato), hidden zero-width characters, padding and small typos.

Set it up with /impersonator, or from the dashboard at lavabot.dev/dashboard/impersonator:

  • /impersonator enable #channel · turn it on and pick the alert channel · /impersonator disable.
  • /impersonator protect add @user · protect a specific person · protect remove · protect list.
  • /impersonator role add @role · protect everyone with a role. Admins and the owner are auto-protected by default.
  • /impersonator whitelist add @user · exempt a member with a naturally similar name.
  • /impersonator threshold 0.88 · sensitivity. Lower catches more, with more false positives.
  • /impersonator action high timeout · auto-action per risk band (none, alert, timeout, kick, ban). Default is alert-only.
  • /impersonator scan · re-check every member now · /impersonator status shows the full config.

Every alert names the suspect, who they resemble, the risk level and the exact techniques detected, with one-click Ban / Kick / Timeout / False positive buttons for your moderators.