Self-bots multitask in ways humans physically can’t. They fire typing
indicators across five channels in the same millisecond, send message
bursts to three rooms before a human could finish a single word. Wiretrip
watches the timing. That’s it.
No message content is read. No privileged intents. The bot never sees
what anyone wrote, only when events arrived and across how many channels.
Every typing event gets timestamped. If the same user triggers activity
in two different channels within the correlation window (150ms by default),
that’s a detection. A score accumulates: faster bursts score higher, more
channels involved scores higher, fresh accounts score higher.
Messages work the same way. And if you mark one channel as a “canary”
(a room that should never see normal traffic), any message there triggers
an immediate flag, no timing analysis needed.
Log – passive. Detections appear in your log channel, nothing gets
banned. Run this for a week first, see what it finds.
Enforce – auto-ban when the score crosses your threshold. 7-day
message purge on ban.
Timeout – auto-mute instead of ban. Duration is configurable.
Hybrid – your moderators get a card with a one-click ban button.
The detection is logged, the human decides.
Every server starts in log mode. Nothing gets banned automatically until
you choose to switch. /wiretrip test posts a synthetic detection so you
can see what the cards look like before anything real happens.
- /wiretrip setup – guided configuration panel
- /wiretrip status – current mode, thresholds, permission check
- /wiretrip history – recent detections with scores
- /wiretrip stats – lifetime pipeline counters
Written in Rust. No external ML services. No message content, no member
list intent. Uses GUILDS, GUILD_MESSAGES, and GUILD_MESSAGE_TYPING only.
Detection runs locally, sub-millisecond response.
