Set up AutoMod — spam, link & hate-raid protection

How to configure FizzyPeak's rule-based AutoMod: sensitivity, spam/link/scam filters, hate-raid and panic protection — per platform. From Creator.

What AutoMod does

AutoMod is FizzyPeak's rule-based chat moderation. Every chat message is scored by a rule engine: hard hits (scams, viewbot spam) are removed instantly, fuzzy signals add up and escalate from warning → timeout → ban. Exempt users (e.g. mods) are never touched.

Important: AutoMod works without AI — the rule engine runs entirely on its own. The AI chat moderator is a separate, optional add-on (see the very bottom) and isn't needed for AutoMod.

AutoMod is available from the Creator plan (€14/month, €12/month yearly). The engine runs across platforms on Twitch, Kick and YouTube. A few special cards (panic mode, hate-raid detection) are Twitch-only and marked accordingly in the dashboard.

Where to find AutoMod

1. In the dashboard, open the Bot tab.
2. Switch to the AutoMod sub-tab there.

Next to it there's also a Moderation tab with simple basic filters (caps, links, words) — that one runs on Twitch only. For cross-platform protection, use the AutoMod tab.

Step 1: choose the scope (per platform)

At the very top of the AutoMod tab it says "Settings apply to:" with a dropdown:

• All platforms (default) — one shared configuration for all platforms.
• Twitch, Kick or YouTube — a dedicated configuration for that platform only.

If you pick a single platform, it first inherits the default configuration. As soon as you change something there, a platform override is created. "Reset to default" undoes the override and follows the shared config again.

You can also turn AutoMod on and off per platform further down (the "AutoMod active on: Twitch / Kick / YouTube" toggles).

Step 2: set the sensitivity (preset)

In the "Sensitivity preset" section you pick a baseline:

• Low — only obvious scams & viewbot spam, minimal false positives.
• Medium — balanced: links, ads, caps & scam packs. Recommended.
• High — aggressive: catches obfuscation, floods & fresh accounts.
• Lockdown — maximum protection for raids and hate-spam events.

As soon as you adjust individual toggles below, the preset switches automatically to Custom.

Step 3: protection toggles (detectors)

In the "Protection toggles" section you enable or disable each detector individually:

• AutoMod master switch — turns the whole engine on/off.
• Block links — filters URLs from non-exempt chatters (clips can be allowed).
• Ad / shop filter — catches "buy followers", shop promos and ad phrases.
• Excessive caps — flags mostly UPPERCASE messages.
• Symbol spam — flags walls of special characters / ASCII art.
• Emote flood — limits the number of emotes per message.
• Repetition / flood — detects copy-paste spam and rapid repeats.
• Raid copypasta nuke — deletes mass-identical messages from a coordinated raid.
• Obfuscation handling — decodes homoglyphs, Zalgo and spaced-out text.
• Fresh-account heuristics — treats brand-new / non-following accounts more strictly.

Scam protection packs

Below that you'll find curated zero-tolerance lists ("Scam protection packs"). Hits are removed silently, without a warning:

• Discord invite spam — unsolicited discord.gg invites.
• Crypto / giveaway scams — fake "Free Crypto" and giveaway bait.
• Art / logo commission scam — mass ads for "cheap logos / art".
• Adult / OnlyFans spam — NSFW self-promo and OnlyFans links.
• Viewbot service spam — blocks the central viewbot-seller domain list.

Step 4: exemptions

In the "Exemptions" section you decide which user tiers bypass AutoMod entirely:

• Moderators bypass
• VIPs bypass
• Subscribers bypass
• Regulars bypass

By default, moderators and VIPs are exempt.

Step 5: block & allow lists

In the "Block & allow lists" section:

• Custom block list — one word or website per line that is always removed (e.g. buy followers cheap or spam-shop.ru).
• Custom allow list — one word, name or website per line that is never blocked (e.g. your own links or regulars). Always wins over the block list.
• Permit duration (seconds) — how long a user cleared via !permit may post one link.

Step 6: bot reply mode

In the "Bot reply" section you decide how the bot communicates when it acts:

• Silent — acts with no chat message.
• Public — posts a short reason in chat.
• Whisper — messages the affected user privately.

Note: scam bans are always silent, regardless of this setting.

Test a message (risk-free)

At the very bottom there's a "Test a message" section. You type an example message, pick the viewer tier (everyone, subscriber, regular, vip, moderator, broadcaster) and click "Test message". The real engine shows you the decision (e.g. "No action — message passes", "Would delete the message", "Would time out", "Would BAN") including the rules that fired — without touching your chat.

Hate-raid & panic protection (Twitch only)

The same tab has cards specifically against hate raids. These features are Twitch-only (Kick/YouTube offer no chat-mode/raid API for this).

Emergency Control — one-click protection:

• Panic Mode — instantly enables follower-only / emote-only / sub-only chat for the set duration.
• Deactivate — ends panic mode again.
• Emergency — switches OBS to a panic scene or stops the stream entirely.

Hate-raid detection ("Automatically detect follow-bot attacks") — if more new followers arrive per minute than the set threshold (follows/min), the bot automatically enables panic mode and warns in chat.

Panic-mode behavior — you configure what happens on !panic or the button:

• Follower-only chat (incl. minimum follower age in minutes),
• emote-only chat,
• sub-only chat,
• plus the panic duration (minutes).

Emergency stream end — defines what !endstream / !emergency triggers: either switch to a panic scene (stream keeps running) or stop the stream entirely.

Matching chat commands for mods

• !permit @user — briefly clears a user to post a link.
• !panic — instantly enables hate-raid protection.
• !endpanic — ends panic mode.
• !emergency / !endstream — triggers the emergency action.

Optional: the separate AI chat moderator

Independent of the rule engine, there's an optional AI chat moderator. It lives not in the AutoMod tab but in the AI Tools tab (together with Auto-Clip), and is off by default.

It's an add-on, not a replacement — the rule-based AutoMod runs fully without it. When enabled, it checks context-sensitively for toxicity/hate speech without wrongly blocking meme language. You control, among other things, strictness (low/medium/high), the action on a hit (warn/delete/timeout), the categories and whether privileged users are exempt.

The AI moderator needs an Anthropic API key set by the platform admin; if none is set, the dashboard shows a matching note and the AI mod stays inactive.

In short: AutoMod = your rules, ready to go from Creator. The AI moderator is the optional cherry on top.

Still have questions? Open the dashboard or message us on Discord in #support.