Dangerous Discord Permissions Explained (and Who Should Have Them)
Eight Discord permissions can damage a server: Administrator, Manage Server, Manage Roles, Manage Channels, Manage Webhooks, Ban Members, Kick Members, and Mention Everyone. A nuke needs at least one, held by a trusted account or an added bot. Use least privilege: keep them with the owner and one admin. Staff moderate through a bot with limits, and anti-nuke punishes abuse.
Why permissions matter more than bots
A security bot can only limit damage from accounts ranked below it. Each dangerous permission you hand out is a promise that the account will never be phished, never leave angry, and never make a mistake at 3 a.m. Discord's audit log records what happened. It does not prevent it. The fewer accounts that can do damage, the less there is to protect.
The dangerous permissions, ranked
| Permission | What it allows | Nuke potential | Who needs it |
|---|---|---|---|
| Administrator | Everything, bypassing all channel overrides | Total | Owner only |
| Manage Server | Add bots, change name, icon, vanity URL, and settings | Adds a malicious bot; defaces the server | Owner and one admin |
| Manage Roles | Create, edit, delete, and assign roles below the user's highest role; grant only permissions the user already has | Hand out existing lower roles that carry dangerous permissions; delete every role below them | Owner and one admin |
| Manage Channels | Create, edit, delete channels | Delete every channel | Owner and one admin |
| Manage Webhooks | Create webhooks that post with any name and avatar | Webhook spam, impersonation | Nobody, usually |
| Ban Members | Ban anyone below the user | Mass ban | Moderators, via a bot with limits |
| Kick Members | Kick anyone below the user | Mass kick | Moderators, via a bot with limits |
| Mention Everyone | Ping @everyone and @here | Mass-mention harassment | Staff who post announcements |
The quiet one
Manage Roles is the one people underestimate. It cannot grant a permission the holder lacks, and it only works on roles below their highest role. The real risk is handing out an existing lower role that carries dangerous permissions, or deleting roles. Security Bot acts on this instantly. It can strip dangerous permissions from an edited role, take back a handed-out admin role, and punish whoever did it.
Permissions that look scary but are fine
- Manage Messages: delete and pin messages. Annoying if abused, not destructive.
- Timeout Members (API name MODERATE_MEMBERS): time members out. The Discord app offers up to one week. Bots and the API can set up to 28 days. Reversible.
- Manage Nicknames, Manage Threads, Manage Events: cosmetic or contained.
- View Audit Log: read-only, and useful for staff.
Security Bot's role triggers treat Manage Messages, Manage Threads, and Manage Nicknames as moderation permissions too. A role that carries them is watched the same way as one with Ban Members.
How anti-nuke watches each permission
Least privilege shrinks the list of accounts that can do damage. Anti-nuke covers the ones that remain. Each dangerous permission maps to a trigger that either counts the action per user (counters reset every 10 minutes) or fires on the first occurrence.
| Permission | Anti-nuke trigger | Fires |
|---|---|---|
| Manage Server | Anti Bot Add (free, on by default); server rename, icon, and vanity URL triggers on Premium | Instantly. The rename, icon, and vanity triggers punish the member; they do not undo the change |
| Manage Roles | Role create and delete limits; Anti Role Permission Change and Anti Admin Role Assign (free, off by default) | Limits when the count reaches them; permission edits and admin role grants instantly, with the dangerous permissions stripped from the role or the role taken back |
| Manage Channels | Channel create and delete limits; channel rename on Premium | When the per-user count reaches the limit |
| Manage Webhooks | Anti Webhook Create (Premium): deletes webhooks just created by a member who is not exempt and punishes the creator; existing webhooks are left alone | Instantly |
| Ban Members | Ban limit for bans made in Discord. While Anti Ban is on, /ban and /softban (and /banid on Premium) through the bot are capped separately at the same limit and refused at the cap | When the count reaches the limit |
| Kick Members | Kick limit; Anti Prune is an instant trigger (free, off by default) | Kicks when the count reaches the limit; prune instantly |
| Mention Everyone | Anti Mention Spam (Premium) plus an AutoMod mention limit | When the per-user count reaches the limit |
Each trigger has its own punishment: kick, ban, or clear roles. The default is kick (ban for prune). Clear roles is a safe choice while you tune. It stops a compromised account without banning a real teammate, and an admin can hand the roles back.
Beast Mode, which is free, adds the actions you give points to into one server-wide score. That way, an attack split across several accounts can still be caught. Its action scores all start at 0, so set them before relying on it.
Bots and permissions
Bot invite links often request Administrator because it is easier than listing twenty permissions. For most utility bots that is far more than they need. A compromised or malicious bot with Administrator is a nuke waiting to happen.
When inviting a bot, untick permissions it does not use. A music bot needs Connect and Speak. A leveling bot needs Manage Roles for its reward roles and nothing else.
Security bots are the exception, because their job is to act on accounts that may hold dangerous permissions. Security Bot requires Administrator. Without it, anti-nuke ignores events and dashboard changes fail. The safeguards:
- Its ban and kick commands are capped per moderator while Anti Ban and Anti Kick are on (they are by default).
- Its moderation commands are written to the moderation log (except
/clearand/raid ban). - Its role position decides who it can act on.
Least privilege for a Discord server
- Owner: the only Administrator. Protect the account with 2FA and a unique password.
- One trusted admin: Manage Server, Manage Roles, Manage Channels. Also 2FA.
- Moderators: no native Ban, Kick, or Manage permissions. Moderation commands through Security Bot with per-user permissions. Their bans and kicks are then capped per 10 minutes while Anti Ban and Anti Kick are on (the default).
- Announcers: Mention Everyone in the announcements channel only, via a channel override.
- Bots: the minimum each one uses. Security Bot's role above all staff roles, other bots below it.
- Everyone: turn on the server's 2FA requirement for moderation, so staff without 2FA cannot use moderation permissions until they enable it.
Suggested setup by server type
The anti-nuke values below are suggested starting points, not defaults. Out of the box the ban and kick limits are 5 per 10 minutes.
| Server type | Native dangerous permissions | Anti-nuke limits | Extras |
|---|---|---|---|
| Small community, under 1,000 members | Owner only, plus one admin with Manage Server, Roles, and Channels | Ban and kick limit 2 or 3, clear roles | One-click or image captcha verification |
| Public gaming server, 1,000 to 20,000 | Owner plus one admin; moderators act through the bot | Ban limit 3, kick limit 3, clear roles | Anti-raid with lockdown on Premium |
| Crypto, NFT, or marketplace server | Owner only; one admin with 2FA, whitelisted per action | Ban limit 3, ban punishment on channel and role delete | Web check with alt blocking, Beast Mode with action scores set, Recovery turned on (Premium) |
The more valuable your server is to an attacker, the more admin work should move behind the bot. The punishment on destructive triggers should get harsher too.
Common permission mistakes
- Administrator on utility bots. The invite asked for it, so it got it. A music or leveling bot with Administrator is one compromised bot token away from a nuke. Re-invite with only what it uses.
- Temporary grants that became permanent. Manage Channels given to a member to build one category and never removed. Audit roles every few months and remove anything nobody can explain.
- Staff roles above the bot. Security Bot can only act on accounts ranked below its role. A Moderator role dragged above it is exempt from every trigger.
- Channel-level overrides. Manage Webhooks or Manage Channels granted on a single channel does not show on the role page. Check channel permissions too.
- Global whitelist entries. Whitelisting the admin role globally turns anti-nuke off for the accounts most worth watching. Whitelist per action.
- 2FA not required. Without the 2FA requirement for moderation, a stolen password is enough to use every permission on the list. With it, a stolen session token still is, so treat 2FA as one layer, not the whole defense.
What Discord provides and what needs a bot
Discord gives you the structure: role hierarchy, channel overrides, the 2FA requirement for moderation, timeouts, Pause Invites, native AutoMod, and an audit log that records everything after the fact. Used carefully, those alone prevent many accidents.
What Discord does not do is count. Nothing native stops an account with Manage Channels from deleting fifty channels in a row. Nothing notices that a role was just edited to include Administrator.
That needs a bot that watches the events as they happen, keeps a counter per user, and acts when the count reaches a limit. It also needs moderation commands that do not require native permissions, so moderators can ban spammers without holding Ban Members.
If the bot goes offline, your Discord permissions are unchanged, and protection resumes when it reconnects. That is why the native setup above matters even with a bot installed. The trust page covers what the bot stores and why it requires Administrator.
If a permission is being abused right now
- As owner, strip the account's roles or ban it. If it is a bot, kick it from Server Settings, Integrations.
- Run
/lockallso members cannot post in your text channels while you look. It only takes away Send Messages. It does not stop changes to roles or channels. - Open the security log for the trigger that fired and the audit log for everything the account touched: channels, roles, webhooks, bots.
- Delete any webhooks or bots it created. If Recovery was turned on before the attack (Premium), restore from the last backup taken before it. You see a preview first. You can also bring back single channels, categories, roles, and emojis from the Recently deleted list.
- Read how to recover after a nuke for the full checklist, then come back to the audit below.
Audit your permissions in ten minutes
- Open Server Settings, Roles and click each role with members. Note every dangerous permission it grants.
- For each one, ask who actually uses it and for what. Remove it if the answer is nobody or if a bot command can do the job.
- Check channel overrides for Manage Channels or Manage Webhooks granted per channel. They are easy to forget.
- Open Integrations and review each bot's permissions. Re-invite over-permissioned bots with a smaller set.
- Run Security Bot's
/scancommand. It lists roles with Administrator, roles with dangerous permissions (and which of those sit above Security Bot), and bot roles with dangerous permissions. The dashboard's Alerts page also flags role hierarchy problems.
Do this with Security Bot
Discord Anti Nuke Bot: Stop nukes with per-user action limits, automatic punishments, owner alerts, and a security log.
Security Bot is a Discord anti nuke bot that watches bans, kicks, channel and role changes, bot adds, prunes, and dangerous permission edits. When an account hits a limit, it is kicked, banned, or stripped of its roles. The server owner gets a DM.
Frequently asked questions
What are the most dangerous Discord permissions?
Administrator, Manage Server, Manage Roles, Manage Channels, Manage Webhooks, Ban Members, Kick Members, and Mention Everyone. Each can be used to damage the server or its members at scale.
Should moderators have Administrator?
No. Give them moderation commands through a bot with limits and logs. Administrator should stay with the owner.
Is Manage Roles dangerous?
Yes, but not because it can create Administrator out of nothing. It can only grant permissions the holder already has, and only on roles below their highest role. The risk is handing out existing lower roles that already carry dangerous permissions, or deleting roles. Limit it to one trusted admin and turn on the anti-nuke triggers for permission changes and admin role grants.
Why does Security Bot need Administrator?
It is required: anti-nuke ignores events without it and dashboard changes fail. With it, the bot can act on compromised staff ranked below its role. Its ban and kick commands are capped per moderator while Anti Ban and Anti Kick are on. Its moderation commands are logged (except /clear and /raid ban). See the trust page for details.
How do I require 2FA for moderators?
Turn on the server's 2FA requirement for moderation in Server Settings. Once it is on, staff without 2FA cannot use moderation permissions until they enable it. It stops attackers who only have a password, not a stolen session token.
Can anti-nuke stop a bot that has Administrator?
Yes, if the bot's role is below Security Bot's role. Adding a bot is an instant trigger, and a bot ranked below Security Bot is subject to the same limits as a user.
Does Security Bot have to be above my staff roles?
Yes. The bot can only act on accounts ranked below its role. Drag it above every staff role in Server Settings, Roles, or those staff are exempt from every trigger.
How this guide was made: drafted with AI assistance, then checked against Security Bot's source code and Discord's official help articles. Spotted something out of date? Tell us in the support server.