Two layers: automatic and manual
- Auto-moderation — Roody analyses every message and applies the actions you've configured (delete, warn, sanction).
- Manual commands — staff sanction, clean up or mute by hand when it's needed.
Auto-moderation filters
Anti-spam (flood)
Detects back-to-back messages from the same user. Configure the number of messages and the time window that counts as spam, and pick the action (warn, delete, timeout).
Anti-links
Blocks external links. You can keep an allowlist of permitted domains (YouTube, Twitch, your own domains…) and choose whether the message is just deleted or the user is sanctioned.
Anti-invites
Blocks invite links to other Discord servers. Optionally allows invites to your own server and to an allowlist of partner servers.
Anti-caps
Detects messages with a high percentage of capitals above a certain length. Configure the percentage and the minimum length for it to kick in.
Anti-duplicates
Detects the same message repeated X times in a row by the same user.
Anti-zalgo
Deletes messages with combining Unicode characters ("zalgo text") that visually break the chat.
Ghost ping
Detects mentions that are edited or deleted afterwards (a technique used to annoy people). It records them in the moderation log.
Each filter has its own configurable action (warn, delete, timeout, kick…) and specific options. Only enable the ones your server needs; having them all on can be too aggressive.
Manual commands
Sanctions
/ban · /unban
Ban or unban a user. /ban takes a reason, which is kept in the log. /unban needs the banned user's ID.
/kick
Kick a user out of the server (they can rejoin with an invite link).
/timeout · /untimeout
Apply or lift a timeout (Discord's temporary mute). Roody accepts durations like 10m, 2h, 1d.
/warn · /unwarn · /warns
A warning system with history. Every warn is recorded and you can look them up with /warns user:@member. /unwarn removes a specific warning.
Channel and messages
/clear
Delete the last N messages (100 at a time at most, and only those under 14 days old).
/slowmode · /unslowmode
Turn slow mode on or off in the channel.
/channel lock · /channel unlock
Lock the channel so only staff can write, or unlock it. Useful during incidents.
Roles and auto roles
/addrole · /unrole
Assign or remove a role from a member straight from Discord (an alternative to clicking through the UI).
/autoroles
Configure roles that are assigned automatically when someone joins the server.
Others
/blacklist
A blacklist of users for the bot's modules (they won't be able to use economy, suggestions, giveaways…).
/antiraid
Manually turn on anti-raid mode or configure its parameters. There's a separate module (/dashboard → Anti-raid) with the detailed configuration.
/verificacion (verification)
Force a user's verification or manage the verification system (configured in its own module). This command isn't translated in Discord.
/say · /reglas (rules)
Make the bot say a message on your behalf, or publish/show the server rules. /reglas isn't translated in Discord.
General module configuration
In /dashboard → Moderation module:
- Enable/disable each filter independently, with its own action.
- Notify the sanctioned user by direct message.
- Embed colour for Roody's notices.
The details of each warn and the thresholds for escalation (3 warns = timeout, 5 = kick, etc.) are configured inside the module, in its sanctions tab.
Permissions Roody needs
- View Channels / Send Messages / Manage Messages (to delete messages that break the filters)
- Kick Members / Ban Members / Moderate Members
- Manage Roles (for auto roles, and to strip the staff role from sanctioned users)
- Manage Channels (for slowmode and lock/unlock)
- View Audit Log (to correlate manual actions)
Recommendations
- Start with the most critical filters (anti-spam, anti-invites) and watch for a while before enabling the rest.
- Always configure the log channel before turning on auto-moderation, so you can see what's being deleted.
- Give the bot's role enough hierarchy. If it sits below the roles you're moderating, the commands will fail.
- Combine warns with escalation: the system sanctions automatically when a user racks up X warnings.
Comments
Share your experience or ask anything about this article.
Join the conversation
Sign in to leave comments and take part in the conversation.
Sign in