Protection Assistant: Roles, Minimum Age and Honeypot Setup

The protection assistant walks you step by step through three layers of protection against spam accounts and raids: new accounts first get no full access, accounts that are too new are turned away, and a honeypot – a bait channel normal members never see – exposes accounts that write or join there. You find it in the setup assistant and in the module marketplace under “Set up protection against spam bots and raids”. It requires the Manage Server permission.

What it cannot do

• No protection catches every spam bot. The honeypot only works if the account actually reaches the channel and does something there.
• Honeypot and minimum age apply to user accounts, including hijacked accounts and selfbots. Real Discord bot applications are deliberately ignored by Kage there.
• The minimum age does not prevent the join. The account joins, then Kage kicks, bans or logs.
• The report is not a functional test. It checks stored settings and calculated permissions, but not a real join and not a real attack.

Two ways

• Set up automatically: Kage reads your server, shows you every planned change and applies it after your confirmation. Existing roles and channels are reused; nothing is deleted and no permission is reset wholesale – only the named permission bits are changed. Changes to existing channels or settings need explicit confirmation. A module is never enabled without your consent. Running the setup a second time creates no duplicates.
• Step-by-step guide: Kage explains every setting in Discord and in the Kage modules (what, where, specifically, why, how to verify, risk). You apply them yourself. A checkmark is only your own note; whether it is true is shown by the check.

Two protection variants

Variant A – with Kage verification. New members first get Unverified and Member after the click in the verification channel. Member cannot see the honeypot. *Remaining gap:* until verified, a new account can still see the honeypot as long as Unverified is not excluded. The report shows you this. If you exclude Unverified, spam accounts without verification no longer reach the bait – the honeypot then mainly catches accounts that are already Member.

Variant B – without verification, with auto role. The member role is assigned right when someone joins; “also bots” stays off. *Remaining gap:* if a normal user does not receive the role (for example because Kage's role is too low), they can see the honeypot. “Exclude bots” does not detect automated human accounts.

How the honeypot works

The decisive setting is View Channel = Deny for Member. Without visibility nobody can normally write. At the same time @everyone must be allowed to see the channel, otherwise an account without a role cannot reach it.

The assistant additionally checks the effective permissions in Discord's calculation order: if another role has Administrator or an allow of its own in the channel, it sees the honeypot despite the block – the report names these roles. Kage cannot evaluate permissions of individual people against their roles and reports them as “manual check required”. The honeypot is only switched on in Kage once the permissions are in place.

Minimum age

24 hours is recommended. You can change the value; 0 switches the check off. With the action “Log only” the minimum age does not protect – it only writes an entry. Accounts that are too new are kicked or banned after the join (see Anti-Raid).

The security report

Every check has an unambiguous status: Passed, Warning (known risk), Failed or Manual check required. Among other things it checks Kage's roles and role position, modules, auto role or verification, minimum age, honeypot channels, channel permissions and conflicts with other roles. It also explicitly names what could not be tested: a real join, triggering the honeypot, the verification flow with a real account, raid mode and user-specific channel permissions. Test these with a second account.

Resuming and errors

Progress is saved continuously; you can stop at any time. “Start over” only forgets the assistant's decisions and changes nothing in Discord. If a Discord call fails (for example because Kage's role position is too low), the assistant reports the step as failed, reads the actual state again and carries on with the remaining steps – without switching on an open bait channel. If a role or module does not fit (for example not in your plan), the step is marked as unavailable; Kage never enables a module unasked.

See also Verification and Auto Roles.