Shape your AutoMod.
Twelve filter categories, editable thresholds, domain and word lists, and channel, role and user exemptions.
Explore the rules →DISCORD MODERATION / ENGLISH FIRST
Less noise.
More good judgement.
Guardify brings configurable AutoMod, practical moderation tools and advisory AI into one Discord bot. Build clear boundaries. Review the evidence. Stay in control.
Add Guardify to Discord ↗Read the setup guide →Independent project. AI review never applies an automatic punishment.
GUARDIFY / CASE DESKILLUSTRATION
01
Catch the pattern.
Repeated messages, invite links, mass mentions.
02
Read the context.
A report is a claim. Available messages are evidence.
03
Local AutoModAdvisory AIGuild-scoped settingsKeep a human in charge.
AI offers observations. Moderators make the call.
THREE LAYERS. ONE WORKFLOW.
Protection you can explain.
Reports with context.
/report user [reason] sends an advisory review of eligible cached messages to moderators. Missing evidence stays missing.
The Moderator’s Desk.
A solo riddle Activity inside Discord. Read a fictional situation, choose a response and learn why it matters.
Meet the game →YOUR SERVER. YOUR RULES.
From the first warning to the next review.
Use the private dashboard to inspect protection, then adjust specific rules with clear slash commands.
Warnings and moderation cases persist locally. Locks preserve the previous everyone-role setting. Sensitive commands check permissions and role hierarchy.
Browse every command →START HERE
A sensible first session
/diagnose — check channel permissions
/log_channel — choose a staff log
/dashboard — inspect protection
/automod filter — tune a rule
/play — open the solo Activity
Default AutoMod enforcement is deletion. Link allowlist enforcement and join-wave timeouts begin disabled.
CLEAR LIMITS
Useful tools, honest boundaries.
No silent AI punishment.
Local AutoMod follows configured rules. AI chat and report advice do not ban, kick or timeout users.
No mystery scanning.
File checks inspect names and extensions. Guardify does not download attachments, visit links or claim to detect all malware.
No invented proof.
Report review uses only eligible messages already cached in the current channel. An allegation is not a finding.