Protection layers

Every layer below is on from the moment you add Mizan. There are no rules to write and nothing to switch on first, and everything here can be loosened or tightened to suit your community.

Chat

The things people actually get hurt by, caught before anyone has to read them.

• Slurs and hate speech, blocked on sight
• Disguised spelling, spacing, and lookalike characters read as what they are
• Abuse spread across several messages, caught as one piece
• Self-harm and encouragement to self-harm, handled with care rather than punishment
• Threats and targeted harassment
• Sexual content aimed at minors
• Personal details posted to expose someone: phone numbers, emails, home addresses, card numbers, IP addresses
• Spam floods, repeated messages, and mass mention pileups
• Your own banned words, phrases, and patterns
• A general profanity filter you can set as strict or as relaxed as your community wants

Links, files, and images

• Phishing pages and sites dressed up to look like Discord, Steam, or a bank
• Domains known for malware and credential theft
• Invites to other servers, allowed or blocked as you prefer
• Images already known to be scams or harmful
• QR codes inside images, checked like any other link
• Adult and gore imagery
• File types that are dangerous to open
• Your own allow list and block list, which always win over ours

Raids and nukes

Raids are about pace and pattern, so Mizan learns what a normal day looks like on your server and reacts to what is unusual for you, not for someone else.

• Join floods slowed and contained without shutting the doors on real members
• Batches of freshly made accounts recognised even when a few older ones are mixed in
• A genuine rush of new members told apart from an attack
• Risky joins sent through verification instead of being turned away
• A trap channel that catches automated accounts before they reach anyone
• Server damage undone: deleted channels, wiped roles, and mass bans put back
• Sudden grabs for dangerous permissions stopped as they happen
• Webhooks being abused to post at speed, shut off
• Slowmode applied on its own when a channel gets out of hand
• A single command to lock the server down, and one to open it again

Names and profiles

Someone who cannot post can still cause harm with a username, so profiles get the same treatment as messages.

• Usernames and nicknames carrying slurs or harassment
• Names built to sit at the top of your member list
• Avatars already known to belong to bad actors
• Custom statuses, thread titles, and attachment names

Tools for your team

• Warn, timeout, kick, ban, and temporary ban
• Bulk message clearing and slowmode
• A full case history behind every action, including automatic ones
• Trusted people, roles, and words that Mizan skips the tunable checks for
• Members can look up their own record without asking a moderator
• A web dashboard for setup, review, and history

Logs, appeals, and reports

• A live moderation log as things happen
• Routine actions collected into a quiet digest instead of a stream of pings
• Member appeals, with the decision resting with you, as they roll out to your server
• Repeat offenders escalated automatically instead of starting fresh each time
• Anything uncertain held for a human to look at
• Reports from members and from staff

Beyond your server

Servers that opt in can share signals about accounts that have already caused damage elsewhere, so an attacker does not get a clean slate on every server they visit. Nothing about your members leaves your server unless you choose to take part.

Banwave

How cross-server signals work, and what is and is not shared.