What Vetox's moderation features actually do

Most bots treat moderation as one feature. Vetox splits it into seven systems, and the split is not marketing — each solves a problem the others structurally cannot.

Understanding the separation is what makes the stack configurable rather than a single on switch.

The seven layers

1. Auto-Moderation — the rule layer

Keyword and pattern filtering with configurable actions per rule: mute, kick, ban, delete. Rules evaluate as messages arrive.

This is the deterministic layer. It is fast, free, auditable, and every case it resolves is one nothing else has to consider. Free on the Starter tier.

2. Protection — the behavioural layer

Where Auto-Mod matches content, Protection matches behaviour: spam bursts, mass mentions, invite posting, link flooding. A single message may be innocuous while ten of them in four seconds are not.

Free on Starter.

3. Anti-Raid — the join layer

Watches join rate rather than messages. A raid is a pattern of arrivals, and no message filter sees it until the damage is underway.

Unlocks at Basic, $4.99/month. Worth comparing: ProBot includes protection from $40/year, and Dyno’s advanced moderation sits inside $49.99/year.

4. AI Auto-Mod — the intent layer

Contextual analysis: tone, meaning, user trust scoring from real-time chat dynamics, and visual scanning for scam imagery. Catches hostility with no filtered word in it, and scams that arrive as pictures.

Tier 3 (Black), with 25 custom AI rules. The AI gate is single — it covers AI Auto-Mod, AI Server Control and the token allowance together.

5. Moderation Tools — the human layer

Where staff act: warnings, timeouts, kicks, bans, case review, and per-member history. Automation handles the mechanical; this is where the contested cases are resolved.

The capability that matters here is queryable per-member history. A third warning is only meaningful if someone can see the first two, and a moderation system whose history lives in moderators’ memories is not a moderation system. Free on Starter.

6. Logging — the record layer

Server events routed to channels you choose, across 148 distinct event types — the largest count of the products we checked, ahead of Sapphire’s 80+, Zeppelin’s 66, and roughly six times Carl-bot’s or ProBot’s ~24. Critically, AI and automated actions land in the same log as human ones, so a member’s timeline is one sequence rather than two systems that do not know about each other.

Free on Starter. Log executor details, webhook delivery and log management unlock at Basic.

7. Dashboard Watcher — the meta layer

Records changes to configuration, not to members. Who altered an automod rule, who changed a permission, who adjusted a limit.

This is among the rarest capabilities in the category. We checked sixteen moderation and security bots against their own documentation: two record who changed their own settings.

Zeppelin keeps config revisions carrying an editing user and a timestamp, visible in its source. Captcha.bot publishes an audit log that shows the value before each change, cannot be switched off, and is free — its export is the paid half. Gaius lists a “Settings” log type among its module actions, which may be the same idea, though its documentation describes itself as several years out of date.

What separates them is scope and retention rather than existence. Captcha.bot audits one verification bot’s settings for 14 days, and has nothing else to audit. Dashboard Watcher spans every system for 90 days, free — which is the difference between knowing who changed the captcha and knowing who changed the automod rule, the permission model and the ticket configuration.

And Zeppelin is one most servers cannot actually use. Its documentation states that access to the hosted bot is “granted on a case by case basis”, the restriction is enforced in its own source, and there is no invite link. Self-hosting means running Docker on Linux.

For the rest, the answer is simply not published — and that is itself the finding. Several of the best-known security bots, Wick among them, publish no list of what they log at all, so a buyer has no way to know whether their configuration changes are recorded before committing to the product. Security Bot, which does publish its log types, has nothing of the kind.

Which is worth pausing on, because this is the layer that matters most precisely where these products are sold. A bot whose entire pitch is protection against a rogue administrator, that does not tell you whether it records that administrator editing its own protection settings, is asking for a kind of trust it has not evidenced.

It becomes essential the moment more than two people can change things. “Why is this rule different from last week” has no answer without it.

Access Control sits across all seven

At Max ($7.99/month), Access Control gates which staff can configure which sections. A moderator can be given the moderation surface without the ability to change billing, AI rules or ticket configuration.

That is the difference between a team of three and a team of twelve. Below it, dashboard access is effectively all-or-nothing, which is why it is the feature that defines the Max tier alongside unlimited limits.

How the layers chain

Discord AutoMod       platform, before delivery — free, configure this first
  ↓
Auto-Moderation       keywords, patterns, per-rule actions
  ↓
Protection            spam, mentions, invites, link floods
  ↓
Anti-Raid             join-rate detection and lockdown
  ↓
AI Auto-Mod           intent, tone, trust score, visual scam analysis
  ↓
Moderation Tools      human review, cases, per-member history
  ↓
Logging               one timeline for automated and human actions
  ↓
Dashboard Watcher     who changed the configuration
  ↑
Access Control        who is allowed to change it at all

Each layer catches what the one above cannot express. Configure upward from the free layers.

What this costs against the alternatives

The comparison that matters is not feature-by-feature but what you get before paying.

Sapphire matches on the filtering half and even beats Vetox on one specific point — keeping every one of its 80+ log types free with no executor gate at all — but it has nothing beyond filtering and logging. Dyno matches on moderation and charges $49.99/year for the levelling and tickets that sit next to it here for nothing.

The stack’s real argument is the last two rows: moderation on its own is a solved problem at every price point. Moderation that shares a permission model, a log routing configuration and a member history with thirty-four other systems is not.

Practical configurations

Small server, no budget. Discord AutoMod plus Vetox Auto-Mod, Protection, Moderation Tools and Logging. Free, and it covers the ordinary cases completely.

Growing server, occasional raids. Add Basic for Anti-Raid at $4.99/month.

Multi-language community. Black for AI Auto-Mod — a wordlist is per-language and detection that works across all 21 supported languages is the only maintainable answer.

Team larger than three. Max for Access Control, so staff hold the surface they need and not the rest.

Related systems

Auto-Mod → AI Auto-Mod → Protection → Anti-Raid → Logging → Dashboard Watcher → Access Control

Adjacent: Applications for vetting members before they gain access, Tickets for handling reports and appeals, and Server Insights for spotting the activity change that precedes a moderation problem.