Stopping scam links, fake giveaways and image scams on a Discord server
By Termi3 min read
Most scams on a Discord server now come from members you know. Their account gets hacked and posts an invite to a "free Nitro" server, a fake Steam login, an @everyone ping about a giveaway, or four screenshots of a crypto casino with no link at all. FiveBot removes these as they land and tells your staff who sent them.
Three guards, one log
FiveBot checks each message as it arrives, without storing what it says. Three modules handle the common scams, and each one has its own exceptions, strikes and punishment.
Anti-link
Block Discord invites only, or every link. Then add exceptions: a streamer role may post Twitch, TikTok and Kick links in the streams channel, and staff categories allow anything. An exception names any mix of roles, channels or whole categories, and domains. GIFs from Tenor and Giphy are off by default and can be allowed through an exception.
Links inside open tickets are allowed by default, so players can still send a clip as evidence. You can turn that off.
Anti-mention
Block @everyone and @here, real pings and attempts alike, and cap how many people and roles one message may tag. You set how many hits in how many seconds lead to a punishment, for example three in 30 seconds before a five-minute timeout, and whether every hit is deleted or only the ones over the limit.
File limit
Image scams carry no link, so anti-link never sees them. The file limit caps how many attachments one message may carry. Two is the default, which still lets a player post a couple of screenshots. Tickets are exempt by default, and so are any channels or categories you list.
What happens on a hit
- The message is deleted.
- A short notice appears in the channel and removes itself after a moment.
- At the limit you set, the sender gets the punishment you picked: none, a timeout, a kick or a ban.
- A log entry goes to your log channel with the author, the channel and the rule. Each module has its own log, or every log can share one channel.
Punishments and logs fire once per member per module per minute, so one compromised account spamming a hundred messages does not bury your log channel. Members with Manage Server always pass.
The rest of the toolkit
- Anti-spam counts messages per member without storing them, clears a flood and can lock the channel while it happens.
- Ghost ping detection catches a mention that was deleted right after it was sent.
- The webhook guard reacts to a webhook that posts @everyone or a burst of messages. If your FiveM server logs to Discord through webhooks, exempt those channels first.
- Web verification sends new members through a Cloudflare Turnstile check on fivebot.ai before they get the member role.
- Delete and edit logs show what a removed or changed message said, for messages sent while the log is on.
Raids and compromised admin accounts have their own tools: join scores, lockdown, anti-nuke and server snapshots. Anti-raid and the ban network covers them.
A starting setup
- Anti-link: block invites, timeout for five minutes, exceptions for your streamer role and your staff category.
- Anti-mention: block @everyone and @here, at most five tags per message, three hits in 30 seconds before a timeout.
- File limit: two files per message, three hits in a minute before a timeout, tickets exempt.
- One log channel for all three, visible to staff only.
Watch the log for a week and loosen whatever catches your own members.
Questions people ask
Is this Discord AutoMod?
No. FiveBot enforces these rules itself, which is what makes the exceptions, strikes and timeouts possible.
Which plan includes it?
All of it, on every plan. Every module is off until you switch it on.
Does FiveBot read and keep my members' messages?
It checks each message as it arrives and keeps nothing. Delete and edit logs are the exception: while they are on, recent messages are held in memory so a deleted one can be shown in the log.