Automoderation

A system for automatically preventing violations in chat. It works in the background, filtering unwanted content and punishing offenders without moderator intervention.

How does it work?

Automoderation has no active commands. It just works. When a violation is detected the bot can:

• Delete the message and notify the user via DM about the reason.

• Apply a timeout to the user.

• Send a notification to moderators in the log channel.

All these actions, as well as exceptions and limits, are configured through the control panel.

Available filters

• Links: Block all links or operate using a whitelist/blacklist of domains.

• Invites: Block invites to other Discord servers.

• Spam: Prevent sending identical or very similar messages.

• Caps Lock: Combat excessive use of capital letters.

• Emoji: Limit the number of emojis in a single message.

• Mentions: Limit the number of user and role mentions, and also block@everyone and @here.

• Zalgo: Filter "corrupted" text.

• Lines: Limit the maximum number of lines in a single message.

• Scam / Phishing: Advanced protection against malicious links (more below).

• NSFW (AI): The neural network automatically detects and blocks adult images, analyzing not only uploaded files but also images by links.

• Toxicity (AI): A unique system that analyzes context, not just keywords. Effectively detects insults, threats and hidden negativity, even if they are disguised.

---

Deep dive: Scam and phishing filter

This is not just a link filter. It's an intelligent system designed to protect your users from fraud, account theft and malicious sites.

Key features

• Each suspicious link is checked against Google’s global database of malicious sites in real time. This provides protection against the latest threats.

• The bot does not simply block everything. It analyzes the message by many indicators. Each indicator adds "danger points." If the total points exceed the threshold 95, the filter triggers.

• The system searches the message text for keywords that create a sense of urgency ("Your account will be suspended!") or excessive gain ("Free Nitro!").

• A combination of heuristics, domain analysis and checking via the Google API provides extremely high phishing detection accuracy (close to 99%), minimizing false positives.

Module configuration

To configure the module use the command /settings automod.

1. General settings: Here you can specify the channel for automoderation logs, a role to ping, as well as global exceptions (roles, channels, categories and specific users, to whom the automoderator will not apply).

2. Filter setup: Select any of the filters to configure its parameters:

   • Enable or disable the filter.

   • Choose actions on trigger (delete, timeout, notify).

   • Set the timeout duration.

   • Set filter-specific limits or exceptions (for example, a domain whitelist for the scam filter).