Auto-Moderation Guide

Configure automated rules for spam, links, invites, blocked words, mentions, caps, and duplicate messages.

Overview

Lyx's auto-moderation system provides granular, rule-based automation. Each rule type can be configured independently with custom thresholds, actions, and exemptions. Rules are evaluated in real-time on every message.

Safe Defaults: Auto-mod is disabled by default. Enable it after configuring rules. Start with conservative thresholds and adjust based on your server's activity.

Enabling Auto-Moderation

/automod enable

Disable with /automod disable. View current config with /automod status.

In the dashboard, use the master toggle at the top of the Auto-Moderation page.

Spam Detection

Detects rapid repeated messages from the same user.

/automod spam max_messages:5 time_window:10 action:delete,warn

max_messages: Messages allowed in window (default: 5)
time_window: Time window in seconds (default: 10)
action: Comma-separated actions (delete, warn, timeout, alert)

Triggers when a user sends more than max_messages within time_window seconds.

Link Filtering

Controls which links can be posted.

/automod links block_all:false allowed_domains:"youtube.com,twitter.com,github.com" action:delete

block_all: true = only allowlisted domains permitted; false = block only non-allowlisted
allowed_domains: Comma-separated domains (no protocol needed)
action: Actions on violation

With block_all:false (default), all links allowed except those matching blocked patterns. With block_all:true, ONLY listed domains allowed.

Invite Blocking

Blocks Discord invite links (discord.gg, discord.com/invite).

/automod invites block_discord_invites:true allowed_invites:"ABC123,my-community" action:timeout,alert

block_discord_invites: Block all Discord invites (default: true)
allowed_invites: Comma-separated invite codes to allow
action: Actions on violation

Word Filters

Custom blocked word list with optional regex support.

/automod words action:add words:"badword1,badword2,(?i)regex_pattern" regex:true

action: add, remove, or list
words: Comma-separated words/patterns
regex: Treat as regex patterns (default: false)

Regex examples: (?i)badword (case-insensitive), \bspam\b (word boundary).

Caution: Regex patterns can have performance impact. Test thoroughly. Use simple word matching when possible.

Mention Limits

Prevents mass mentions in a single message.

/automod mentions max_user_mentions:5 max_role_mentions:2 action:delete,warn

max_user_mentions: Max @user mentions (default: 5)
max_role_mentions: Max @role mentions (default: 2)
action: Actions on violation

Caps Detection

Detects excessive capitalization in messages.

/automod caps min_length:10 max_percentage:70 action:delete

min_length: Minimum message length to check (default: 10)
max_percentage: Max % caps allowed (default: 70%)
action: Actions on violation

Only checks messages longer than min_length. Calculates uppercase letter percentage.

Duplicate Detection

Detects repeated similar messages (copypasta, spam).

/automod duplicates time_window:60 similarity:90 action:delete

time_window: Seconds to look back (default: 60)
similarity: Similarity threshold % (default: 90)
action: Actions on violation

Compares new messages to recent messages using similarity algorithm.

Action Types

Each rule can take multiple actions. Available actions:

Action	Description
`delete`	Delete the triggering message
`warn`	Issue a warning to the user (creates case)
`timeout`	Timeout user (requires duration config)
`alert`	Send alert to staff log channel

Timeout duration is configured per-rule in the dashboard, or uses server default.

Exemptions

Exempt specific roles or channels from all auto-mod rules.

/automod exemptions type:role action:add target:@Moderator
/automod exemptions type:channel action:add target:#bot-commands

type: role or channel
action: add, remove, list
target: Role mention or channel mention

Exempted users/channels bypass ALL auto-mod rules. Use for moderators, bots, or testing channels.

Log Channel

Set a dedicated channel for auto-mod action logs.

/automod logs channel:#automod-logs

Logs include: rule triggered, action taken, user, channel, matched content (truncated).

Testing Rules

Before enabling rules in production:

Configure rule with action:alert only
Test in a controlled channel
Monitor alerts in log channel
Adjust thresholds as needed
Add final actions (delete, warn, timeout)
Enable auto-mod globally

The dashboard provides test controls for each rule with sample messages.

Dashboard Builder

The Auto-Moderation page (/dashboard/{guildId}/automod) provides a visual builder:

Rule cards with enable/disable toggles
Threshold sliders and inputs
Action checkboxes per rule
Exempt role/channel multi-selectors
Test message input with live evaluation
Recent automod events table

All changes save instantly via API with loading/error states.

Common Issues

False positives on spam detection

Increase max_messages or time_window. Exempt active channels like #bot-commands. Consider higher thresholds for large servers.

Links not being blocked

Check: 1) Rule is enabled, 2) Auto-mod is globally enabled, 3) Channel/role not exempted, 4) Domain matching is correct (subdomains included).

Word filter not catching variations

Use regex for variations: (?i)badword for case-insensitive, badwo?rd for optional letters. Test with /automod words action:list.

Actions not executing

Ensure Lyx has required permissions: Manage Messages (delete), Moderate Members (timeout/warn), View Channel/Read History. Check role hierarchy.