Guide · October 19, 2026 · 5 min read

Discord honeypot: a trap channel against hacked accounts

A hacked account doesn't read your channels: it posts everywhere, starting at the top of the list. A trap channel exploits that reflex. This guide explains how it works, why most honeypots punish the wrong people, and how to set one up that repairs instead of sanctioning.

The idea: a channel nobody should use

A honeypot is a channel placed at the very top of the list, with a name and a warning that leave no doubt: do not type here. A human reads the warning and moves on. A hacked account or a selfbot reads nothing: the program driving it posts the same scam in every channel it can write in, starting with the first one.

The trap channel thus becomes a detector: whatever lands there is, in the vast majority of cases, an account no longer controlled by its owner.

Why hacked accounts are the real problem

A hacked account is more dangerous than an attacker's account: it is known. Its messages come from an established, sometimes well-liked member, and others click. That is how fake Nitro, fake "game tests" and fake login pages spread, as described in the guide on scams: each victim becomes the next sender.

Link and image filters stop part of these messages. The trap channel spots the account, from its very first move, even when the message contains nothing known.

The mistake of ordinary honeypots: banning everyone

Most honeypots ban anyone who writes in the channel. The result: the new member asking "what is this channel?" gets banned, and the owner of the hacked account — the real victim — loses access to the server without understanding why, with their scam messages still visible everywhere.

A good honeypot must therefore do two things the others don't: tell a driven account from a curious human, and repair rather than punish.

Telling a driven account from a curious member

OriusBot judges every message in the trap channel on concrete clues before deciding anything.

One strong clue is enough to conclude; so are two weak ones. Without any clue, the message is judged human. And since a driven account hits the trap before other channels, OriusBot watches for three more seconds before believing it is a human: time enough for a program to give itself away by posting elsewhere.

Team members who can manage messages are never sanctioned: their message is deleted with a "test passed" note. You can therefore check the trap without any risk.

Repairing instead of punishing

By default, a driven account gets a softban:

1. Its owner is warned by direct message before the sanction (afterwards, they would no longer share a server with the bot): their account seems hacked, and here is how to get it back — change the password, enable two-factor authentication, log out unknown devices.
2. Its messages are deleted from the whole server, not just from the trap: the scam disappears from the channels where it had already been posted.
3. The ban is lifted immediately, and the direct message contains a single-use invite, valid for seven days, to come back once the account is secured.

A curious human sees their message deleted and receives an explanation. No sanction the first time; a ten-minute timeout only if they do it again the same day.

Every decision is explained in the log channel, clue by clue, with an Undo sanction button. A mistake is fixed in one click.

The Orius Network: one catch protects other servers

A hacked account doesn't target a single server: it visits every one it belongs to. When it is caught on a server protected by OriusBot, it is flagged for 24 hours to the others: its arrival is reported to the team, and its images get priority analysis.

Only that fact is shared. Neither the name of the original server, nor the message, nor the clues leave it. An undone catch immediately stops being flagged, and every server can leave the network.

Setting up the trap channel

1. Type /setup-honeypot, or open the Honeypot page of the web panel.
2. Click Create a trap channel: OriusBot creates a ⛔・do-not-type-here channel at the very top, open to everyone, with a warning. You can also use existing channels (up to five traps).
3. Choose the response to driven accounts: softban (recommended), ban, kick or a 24-hour timeout.
4. Choose the response to humans: delete and warn (recommended), ten-minute timeout, or delete only.
5. Pick a log channel to follow the catches.

OriusBot needs the Manage Channels, Manage Messages and Ban Members permissions, with a role placed above the members to protect.

Placing the trap well

• At the very top of the list: that is where programs start.
• Visible and writable for @everyone: a trap nobody can see catches nobody.
• A name that clearly says not to write: it protects humans, not bots.

Frequently asked questions

Will a member who writes by mistake be banned?

No. Without any clue of a driven account, their message is deleted and they receive an explanation. They only get a ten-minute timeout if they do it again the same day.

Does the honeypot replace the anti-spam?

No, it complements it. The anti-spam judges what is posted; the trap channel spots an account no longer controlled by its owner, often before it even posts elsewhere.

What does the victim of a hacked account see?

A direct message telling them their account seems hacked, the steps to recover it, and an invite to come back. Their scam messages have disappeared from the server.

Does my server share information with others?

Only the fact that an account was caught, for 24 hours, and only if you take part in the Orius Network. No server name and no message leaves your server.

Read next : Discord scams, Protecting your server from raids, Security checklist.