AI moderation triage that understands what people are saying — not just which words they type. Flags trouble. Stays out of the way.
Every message runs through ModDog's detection pipeline. Only what needs attention reaches your mods.
One click. ModDog joins your server and starts watching monitored channels. No config files, no endless setup.
Our classifier reads every message — toxicity, spam, harassment, self-harm. It catches what keyword filters miss.
Flagged messages land in your private mod-queue with category, confidence score, and context. Your mods decide — ModDog never auto-punishes.
ModDog watches your server's pulse — not just individual messages.
Detects mass-join raids and channel/role deletion bursts by tracking event velocity in a sliding window. Alerts your mod team before the damage spreads. RAM-only — no data persisted.
When a nuke is confirmed, ModDog resolves the actor from the audit log and strips their roles — dropping them into a locked-down quarantine role. Reversible by design.
Encrypted snapshot of your server's ModDog config and alert history. AES-256-GCM sealed, checksum-verified, restorable in one command. Your settings survive restarts and migrations.
Type a message — it runs through ModDog's real pipeline: L0 rules → the AI judge.
Type something a mod might need to see, or normal chat to see it pass.
One plan covers all your servers. No per-server pricing.
Pricing coming soon.
ModDog classifies messages in memory and doesn't keep your chat history. Settings and the audit log are encrypted at rest (AES-256-GCM), and every user identity is a per-server salted hash — un-linkable across servers. Your community's conversations stay yours.
"ModDog catches stuff our mod team would've missed. The contextual detection is night-and-day compared to the keyword filter we were using."— Discord server mod, 500+ member community