Discord server nuked? Your first hour, in order
The channels are gone and your phone won't stop buzzing. Here's the sequence that gets the server back without letting the attacker wipe it a second time.
Short answer
Lock the attacker out before anything else: secure the owner account, remove unknown roles, webhooks and integrations, and rotate bot tokens. Record the audit log, then restore a RestoreCord snapshot from before the first destructive change, pull verified members back and run Restore Roles.
Don't press restore yet
The instinct is to rebuild immediately, because every minute the server sits empty feels like members drifting away. Resist it for a few minutes. If whoever did this still holds Administrator, a bot token or the owner login, they can watch you rebuild and wipe it again. Worse, a restore overwrites the very evidence you need to work out how they got in.
Nukes rarely come from a stranger breaking in. In practice it's a phished owner or admin account, a leaked bot token, or a staff member who decided to burn the place down. All three have real permissions, which is why containment comes first.
Minutes 0 to 15: close every door
- Owner account. Change the password from a device you trust, confirm two-factor authentication, sign out sessions you don't recognise and secure the email address behind the account.
- Admin roles. Strip Administrator and Manage Server from anyone you aren't certain about, even temporarily. You can hand them back later.
- Webhooks, integrations and apps. Remove anything the team didn't add. Attackers like leaving a webhook behind so they can spam the rebuilt server.
- Bot token. If there's any chance it leaked, reset it in the Discord Developer Portal and paste the new token into the bot's credentials in RestoreCord before running anything else.
Never paste a new token or password into a ticket, DM or staff chat while you're cleaning up. If one of those accounts is compromised, you've just handed it over again.
Minutes 15 to 25: write down what happened
Open the audit log and screenshot the destructive entries before you change anything else. Note the time of the first malicious action in UTC. That timestamp decides which snapshot is safe, and it's what Discord's Trust & Safety team will ask for if you report the account.
A short, plain-text note is enough. Keep it somewhere the attacker can't reach, such as a document outside Discord:
First destructive action: 21:04 UTC (channel delete, by @mod_account)
Owner secured: 21:09 UTC
Access paths closed: admin roles, 2 webhooks, bot token reset
Last clean snapshot: nightly, taken before 21:04
Restore started: 21:18 UTC, into the same server
Still to do: member pull, Restore Roles, staged reopenPicking the right snapshot
The newest snapshot isn't automatically the right one. Match the situation to the restore point:
| What you see | Which snapshot | Why |
|---|---|---|
| Everything deleted in one burst | The newest one taken before the burst | It holds the server exactly as it was |
| Slow sabotage over several days | One from before the first odd change | Later snapshots already include the damage |
| A scheduled snapshot ran mid-attack | The one before it | Scheduled snapshots copy whatever exists at that moment |
RestoreCord's backup previewer lets you browse a snapshot's channels and messages before restoring, so you can confirm you picked the version you want.
Snapshot captured before the permission redesign.
Roles and overwrites look right in the preview.
Backup completed — 42 channels, 18 roles, 6,300 messages.
Restoring the structure
Restore into the same server if you still control it. It keeps the server ID and every invite link you've shared. Drag the bot's role above everything it has to recreate, confirm the server RestoreCord shows is the right one, choose how many messages to bring back and start. Channels, categories and roles come first, then message history.
There's an option to clear the destination first. It's destructive and wipes whatever survived, so save it for a deliberate clean rebuild. Watch the task while it runs and note skipped objects or permission errors rather than launching a second restore on top.
Getting your members home
Members aren't stored in a snapshot. If the attacker banned people, unban them, then start a member pull into the repaired server. Everyone who verified through your RestoreCord bot before the incident is eligible, and RestoreCord sets no cooldown between pulls on any plan. Once they're back, give them their roles from the Pending Role Restore banner, covered in getting roles back after a restore.
Reopen in stages
Check that private channels are still private, test verification with a spare account, then let staff in, then a small group, then everyone. Take a fresh snapshot once things are calm. More on the whole process on the RestoreCord Recovery home page. If the attack came from inside your team, read what to do when an admin goes rogue, and if a bot token was involved, reset it the safe way.
Nuke response checklist
Take the owner account back
Change the password from a device you trust, check two-factor authentication, sign out unknown sessions and secure the email behind it.
Cut every unknown access path
Remove admin roles, webhooks, integrations and apps you don't recognise, and reset any bot token that might have leaked.
Record the evidence
Screenshot the audit log and note UTC times of the first destructive action before you change anything else.
Restore, then reopen in stages
Restore the snapshot, pull members, restore roles, then let staff in first, a small group second and everyone last.
After-nuke questions
Next time, be the server that bounces back
Keep a recent RestoreCord snapshot and a verified member list, and a nuke turns into a restore instead of a farewell post.