# Toxicity Prevention

Automatically detect and action toxic language, slurs, harassment, and hostile messages.

## How It Works

The toxicity filter analyses every message sent in monitored channels against a dictionary of known toxic patterns and contextual language models. When a message exceeds the configured sensitivity threshold, Exorion applies your chosen action.

## Sensitivity Levels

| Level | Description | Typical Use |
| --- | --- | --- |
| Low | Only catches explicit slurs and severe toxicity. | Public communities that want minimal interference. |
| Medium | Catches moderate insults, targeted harassment, and hate speech. | General-purpose servers. Recommended default. |
| High | Catches subtle toxicity, passive-aggressive language, and borderline content. | Strict communities, brand servers. |
| Strict | Flags almost any potentially negative language. | Child-safe or heavily moderated environments. |

## Actions

When a message is flagged, Exorion can perform one or multiple actions:

- **Delete Message:** Silently removes the flagged message from the channel.

- **Warn User:** Sends a DM or inline warning. Logged to the infraction system.

- **Timeout User:** Applies a Discord timeout for a configured duration.

- **Ban User:** Immediately bans the user. Reserved for severe or repeat offences.

## Channel & Role Exemptions

You can exempt specific channels (e.g. staff-chat) or roles (e.g. Moderator) from the toxicity filter. Navigate to **Dashboard -> Auto Moderation -> Toxicity** and scroll to the exemptions section.

> Setting the sensitivity to **Strict** may cause false positives. We recommend starting at Medium and adjusting based on your community.