Auto moderation
Spam protection with image recognition, advertising and scam filters, cross-channel spam and strikes, plus filters for mentions, spam and blocked words.
What is it for?
Protects your server before your team has to step in. It reads the text inside images to catch scam ads, blocks advertising as well as job and crypto scams, removes the same message posted in several channels and filters invites, links, mass mentions and blocked words.Set it up in 3 steps
- 01Switch the module on. Spam protection and image recognition are already on by default.
- 02Choose the punishment: strikes (1 day, 7 days, then a ban), a timeout or just deleting.
- 03Add exceptions: roles and channels that are never checked, channels where links are allowed and channels where certain roles may advertise. Save.
TipEvery case appears in your log channel with buttons: lift the timeout or ban with one click, or report a false detection to us so we can improve it.
All settings explained
Switched on and off with the switch at the top of the tab.
Spam protection
- Stop cross-channel spam
- Removes the same (or almost the same) message when it lands in two channels within 5 minutes, both copies.
- Block advertising
- Recognises self-promotion like "I offer cheap scripts, DM me", paid services, remote job, KYC and account offers and contacts via Telegram or WhatsApp.
- Block scam messages
- Fake giveaways, casino and crypto bonus codes, fake payouts. Scored like the image recognition, so normal chat about games stays untouched.
- Punishment for spam, scam and advertising
- Applies to invites, links, cross-channel spam, advertising, scam messages and scam images. Strikes: 1st time 1 day timeout, 2nd time 7 days, 3rd time ban. Every case can be lifted with one click in the log.
Image recognition
- Detect scam images (text recognition)
- Reads the text inside every image and removes fake giveaways, casino and crypto ads. Game screenshots stay untouched.
- Stop image spam across channels
- The same image in a second channel within 9 minutes, or images in three channels within 2 minutes. All copies are removed.
- Shared scam image blocklist
- A scam image recognised on any AgencyMod server is removed everywhere instantly, without waiting for the text recognition.
Invites and links
- Block invite links
- Removes invites to other Discord servers.
- Block links
- Removes all links except the allowed domains below.
- Allowed domains
- For example youtube.com. Subdomains like www.youtube.com are allowed too.
- Channels where links are allowed
- For example a media or clips channel. Invites stay blocked there.
Advertising channels
- Advertising channels
- Channels or whole categories where advertising is allowed, for example #self-promo.
- Who may advertise there
- Only these roles. Empty means everyone. The filters don't check them in these channels.
Mentions
- Limit mentions
- Removes messages that mention too many members or roles at once.
- Maximum mentions per message
- @everyone and @here count as five.
Spam and capitals
- Stop fast spam
- Too many messages from one member in a short time.
- Number of messages
- How many messages count as spam…
- Within seconds
- …within this many seconds.
- Block repeated messages
- The same message three times within 30 seconds.
- Limit capital letters
- Removes messages written mostly in capitals (from 12 letters on).
- Maximum share of capitals (%)
- From this share on the message is removed.
Blocked words
- Block words
- Removes messages that contain one of the blocked words.
- Blocked words
- Whole words, upper and lower case don't matter.
Light violations
- Punishment for light violations
- Mentions, fast spam, repeated messages, capital letters and blocked words. The message is always deleted, you can also warn the member or time them out.
- Timeout length (minutes)
- Used when Timeout is chosen as the punishment.
- Short notice in the channel
- Tells the member why the message was removed. The notice disappears after a few seconds.
Exceptions
- Ignored roles
- Members with these roles are never filtered.
- Ignored channels
- Channels or whole categories without filters.
AI protection
- AI scam detectionPremium
- AgencyAI also recognizes scam and phishing messages that no word list catches.
- AI filter for insults and hatePremium
- AgencyAI recognizes insults, hate speech and threats in every language, even when they are written in a disguised way.
Something unclear? Open a ticket at help.agencyg.de in the AgencyMod category or ask in our Discord. Privacy policy and terms