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

# Rules

> Understanding moderation rules and rule-based filtering

## Overview

Rules define content moderation criteria in PixelPatrol. Each rule evaluates specific aspects of submitted content, and rules work together within [Rule Groups](/concepts/rule-groups) to provide comprehensive content filtering.

**Important**: All rule management is done through the dashboard interface - no API access is available for creating, editing, or managing rules.

## How Rules Work

### Rule Group Logic

Within each rule group, **ALL rules must pass** for content to be approved. If any single rule fails, the content is flagged for moderation. This ensures comprehensive content filtering without compromise.

**Example**: A "Profile Photos" rule group with 3 rules:

* ✅ No explicit content (passes)
* ✅ Must have clear face (passes)
* ❌ **No children in photos** (fails)
* **Result**: Content is rejected because one rule failed

### Rule Types

<CardGroup cols={2}>
  <Card title="Requirements" icon="circle-check">
    Content must satisfy this condition:

    * Must have clear face
    * Must be high quality
    * Must match profile
  </Card>

  <Card title="Prohibitions" icon="ban">
    Content must avoid this condition:

    * No explicit content
    * No violence/weapons
    * No children
  </Card>
</CardGroup>

## Available Rule Templates

PixelPatrol provides 16 pre-built rule templates optimized for different moderation needs:

### Content Safety

* **Explicit Content** - Blocks NSFW/adult content
* **Violence/Weapons** - Prevents violent imagery
* **Drugs/Alcohol** - Filters substance-related content
* **Offensive Gestures/Symbols** - Blocks inappropriate symbols

### Image Quality

* **No Clear/Focused Face** - Ensures face visibility
* **Unclear/Blurry Image** - Maintains quality standards
* **Fake/Generated Image** - Detects AI-generated content
* **Screenshot Detected** - Prevents screenshot uploads

### Content Compliance

* **Contains Children** - Protects child safety
* **Multiple Focal Points** - Ensures single-person focus
* **Celebrity Detected** - Prevents celebrity photos
* **Excessive Text** - Limits text overlays

### Privacy & Identity

* **Consent/Privacy Violation** - Protects user privacy
* **Offensive Text/Images** - Blocks inappropriate overlays
* **Gender Mismatch** - Validates gender matches profile (requires advanced parameters)

### Custom Rules

Create completely custom rules for specific business requirements not covered by templates.

## Rule Configuration

Each rule can be configured with:

* **Name**: Descriptive identifier for the rule
* **Description**: Explanation of what the rule does
* **Evaluation Type**: Requirement or Prohibition
* **Confidence Threshold**: How confident AI must be (50%-100%)
* **Active Status**: Enable/disable without deleting
* **Advanced Parameters**: Additional metadata requirements

### Confidence Thresholds

Set how confident the AI must be in its evaluation:

* **High (85-95%)**: Fewer false positives, more false negatives
* **Medium (70-85%)**: Balanced approach (recommended starting point)
* **Low (50-70%)**: More false positives, fewer false negatives

## Advanced Parameters

Some rule templates support advanced parameters that use metadata from your media submissions for more precise moderation.

### Gender Matching Example

The "Gender Mismatch" rule requires gender information from media metadata:

<CodeGroup>
  ```json Advanced Parameters Schema theme={null}
  {
    "type": "object",
    "required": [
      "gender"
    ],
    "properties": {
      "gender": {
        "type": "string",
        "description": "User's stated gender (e.g., male, female, non-binary)"
      }
    }
  }
  ```

  ```json Media Submission theme={null}
  {
    "api_key": "site_xxxxxxxxxxxxxxxxxxxx",
    "content_url": "https://example.com/profile-photo.jpg",
    "app_media_id": "user-123-profile",
    "metadata": {
      "gender": "female",
      "user_id": "user-123"
    }
  }
  ```
</CodeGroup>

The AI validates that the photo matches the stated gender, helping ensure profile authenticity.

## Industry Use Cases

### Social Media Platform

<AccordionGroup>
  <Accordion title="Profile Photos Rule Group">
    * No Clear/Focused Face (requirement)
    * Explicit Content (prohibition)
    * Contains Children (prohibition)
    * Celebrity Detected (prohibition)
    * Gender Mismatch (prohibition) - with gender parameter
  </Accordion>

  <Accordion title="User Content Rule Group">
    * Explicit Content (prohibition)
    * Violence/Weapons (prohibition)
    * Offensive Gestures/Symbols (prohibition)
    * Drugs/Alcohol (prohibition)
  </Accordion>
</AccordionGroup>

### E-commerce Platform

<AccordionGroup>
  <Accordion title="Product Photos Rule Group">
    * Screenshot Detected (prohibition)
    * Unclear/Blurry Image (prohibition)
    * Excessive Text (prohibition)
    * Offensive Text/Images (prohibition)
  </Accordion>

  <Accordion title="Review Images Rule Group">
    * Fake/Generated Image (prohibition)
    * Consent/Privacy Violation (prohibition)
    * Explicit Content (prohibition)
  </Accordion>
</AccordionGroup>

### Dating Application

<AccordionGroup>
  <Accordion title="Main Profile Photos">
    * No Clear/Focused Face (requirement)
    * Multiple Focal Points (prohibition)
    * Explicit Content (prohibition)
    * Contains Children (prohibition)
    * Gender Mismatch (prohibition) - with gender parameter
    * Fake/Generated Image (prohibition)
  </Accordion>
</AccordionGroup>

## Creating Rules via Dashboard

1. **Navigate to Rule Groups**: Sites → \[Your Site] → Rule Groups
2. **Select Rule Group**: Choose the group to add rules to
3. **Click "Add Rule"**: Opens the rule creation dialog
4. **Select Template**: Choose from pre-built templates or "Custom"
5. **Configure Settings**:
   * Set evaluation type (requirement/prohibition)
   * Adjust confidence threshold
   * Add any advanced parameters
6. **Test & Save**: Use test interface to validate before saving

## Best Practices

### Rule Organization

* **Group by purpose**: Keep related rules together
* **Limit group size**: 3-7 rules per group for optimal performance
* **Name clearly**: Use descriptive names that explain the rule's purpose

### Threshold Tuning

* **Start at 85%**: Begin with higher confidence and adjust down if needed
* **Test with real content**: Use actual examples from your platform
* **Monitor false positives**: Review flagged content regularly
* **Adjust gradually**: Make small threshold changes (5-10%)

### Advanced Parameters

* **Validate availability**: Ensure your app provides required metadata
* **Handle missing data**: Consider fallback behavior
* **Privacy first**: Only collect necessary information
* **Document requirements**: Clearly communicate metadata needs

## Monitoring Performance

Track rule effectiveness through the dashboard:

* **Trigger Frequency**: Which rules activate most often
* **Confidence Distribution**: How confident AI is in decisions
* **False Positive Rate**: Rules that may need adjustment
* **Processing Time**: Rule evaluation performance

<Tip>
  **Pro tip**: Review your moderation logs weekly to identify patterns and optimize rule configurations. Rules that trigger too frequently or rarely may need threshold adjustments.
</Tip>

## Limitations

* No composite rules with OR logic - all rules must pass
* No variables or dynamic values in rule configuration
* No API access - all management through dashboard only
* Rules evaluate independently - no dependencies between rules

## Related Concepts

* [Rule Groups](/concepts/rule-groups) - How rules are organized
* [Sites](/concepts/sites) - Where rule groups are assigned
* [Moderation](/concepts/moderation) - Overall moderation process