AutoMod
LiveSix filters, one sanction per filter, no permanent surveillance.
The filters
Each filter is turned on alone and carries its own sanction. A server that only wants to block invites does not have to put up with an anti-spam it never asked for.
| Filter | What it spots |
|---|---|
spam | Too many messages in a sliding window, per channel and per member |
mentions | Too many distinct mentions in one message, @everyone included |
invites | Invites to another server, except codes declared as partners |
links | Links whose domain is not on the allowed list |
caps | Messages mostly in capitals, beyond a given length |
words | Banned words, compared without accents or case |
The sanction
The message is deleted in every case — leaving the content in place while sanctioning its author would make no sense. What follows depends on the filter.
| Action | Effect |
|---|---|
DELETE | Deletes, without opening a case |
WARN | Opens a warning case, which counts towards thresholds |
TIMEOUT | Silences for the duration set on the filter |
KICK | Kicks |
BAN | Bans |
From WARN upwards, the intervention opens a case just like a manual command, with the origin AUTOMOD. It therefore shows up in /history and counts towards warning thresholds.
Who escapes the filters
Exempt roles and channels are skipped before any analysis. Axaro also checks its own hierarchy: if it cannot sanction the person, it logs instead of trying an action bound to fail.
- Bots are never analysed.
- Roles listed as exempt go through unchecked, staff included.
- A test channel can be excluded entirely.
Common workarounds
Before comparison, the text is brought back to its simplest form: accents removed, case dropped, punctuation replaced by spaces. The usual trick of writing a word letter by letter separated by dots does not get through.
ass will not catch class — that is intended, and it avoids half of the false positives of word lists.