> ## Documentation Index
> Fetch the complete documentation index at: https://docs.headlesslabs.xyz/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Content safety

> Configure Sentinel message filters and distinguish their exemptions from Discord's native AutoMod rules and the Anti-Nuke whitelist.

## Three separate controls

Headless Sentinel analyzes messages and events according to server configuration. Native AutoMod rules belong to Discord and are managed in Discord's settings. Anti-Nuke can record changes that weaken those rules; that detection does not create rules or replace message filters.

In the Dashboard, open **Seguridad → AutoMod**. This page manages Sentinel filters and their content exemptions. Message analysis runs only while Sentinel is enabled; Monitor mode or Simulation records detections without applying punishments.

> [!NOTE]
>
> Content exemptions are separate from Anti-Nuke Admin access, the Anti-Nuke whitelist, and native Discord permissions.

## What Sentinel analyzes

Configuration can limit long messages, rapid repeats, and bursts of real `@here` or `@everyone` mentions. Separate engines also cover spam, invites, media, mentions, and identity; each response depends on Sentinel and the relevant engine being enabled, the server mode, and the bot's effective permissions.

For the message-spam filter, the saved punishment can be `delete`, `review`, `timeout`, `jail`, `kick`, or `ban`; the initial value is `delete`. In Monitor or Simulation, Sentinel records detections without applying punishments.

> **Message-analysis bounds**
>
> * Maximum message length: 400–2,000 characters; the initial value is 1,800.
> * Repeated-message window: 2–60 seconds; the initial value is 10 seconds.
> * Identical short-message repeats: 3–12; the initial value is 5.
> * Global-mention bursts: threshold of 2–12 within a 2–60-second window; initially, two mentions in 10 seconds.

> **Content exemptions**
>
> The selector handles up to 25 members per operation, excludes bot accounts, and keeps up to 120 unique IDs per list and server. `Long messages` exempts those members from content analysis; `@here` and `@everyone` profiles are separate permissions for each real broadcast mention; `Full announcements` combines all three scopes.

> **Discord AutoMod**
>
> Headless does not expose a public command for server staff to create or edit native Discord rules. Manage those in Discord. The Anti-Nuke `automod_weakening` detection watches for changes that weaken them, but it does not create rules or moderate messages.

> [!TIP]
>
> One real `@here` or `@everyone` mention is not treated as a burst. The engine requires at least two mentions before that threshold can trigger.

## Access and commands

These are server-only commands and require the server owner or Headless owner (configured in Headless). An Anti-Nuke Admin does not gain access to these actions through that role. `.automod` and `.automod content-exemptions` open the panel in the current channel; use a private staff channel so the exemption controls are not exposed to members.

```headless-command Syntax theme={null}
.automod
```

```headless-command Example theme={null}
.automod
```

```headless-command Syntax theme={null}
.automod content-exemptions
```

```headless-command Example theme={null}
.automod content-exemptions
```

```headless-command Syntax theme={null}
.automod announcement-role [role] [action=toggle]
```

```headless-command Example theme={null}
.automod announcement-role @Staff on
```

> [!WARNING]
>
> A message exemption does not automatically authorize `@here` or `@everyone`. Broadcast mentions have exact, separate permissions; an exemption also does not bypass Sentinel's other engines or Anti-Nuke protection.

## Safe configuration

> [!TIP]
>
> Start in Monitor or Simulation and review detections before allowing automatic
> responses.

**1. Review Sentinel in the Dashboard**

Open **Seguridad → AutoMod** and check whether Sentinel and the relevant engines are enabled. Review the server mode before changing thresholds.

**2. Observe before applying punishments**

Keep Monitor or Simulation on while reviewing detections, and confirm the bot still has the permissions and role hierarchy it needs.

**3. Add precise exemptions**

As the server owner or Headless owner, open `.automod content-exemptions` in a private channel and select only the members and scopes needed. For a role, run `.automod announcement-role @Role on`; remove it with `off`.

**4. Manage native rules in Discord**

Create or change Discord's own rules in the server's AutoMod settings. Do not confuse that rule set with Sentinel content filters.

> [!NOTE]
>
> Message and role exemptions only skip the stated content analysis. The `@here` and `@everyone` permissions are exact, user-scoped exceptions; they do not create a general exemption.

## Related pages

> **[Anti-Nuke](/docs/en/security/anti-nuke.md)**
>
> Review how mode, Simulation, modules, and permissions determine a security response.

> **[Moderation](/docs/en/security/moderation.md)**
>
> Read about moderation actions, access, and server protections.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.