Set up automated content filtering to keep your hub safe and welcoming. Learn pattern matching, filter actions, and best practices.

Content Filtering & Safety

Automated content filtering is your first line of defense against inappropriate content. This guide covers setting up effective filters, understanding pattern matching, and implementing a layered safety approach.

What You'll Learn

How to configure automated content filtering
Pattern matching techniques for effective filtering
Choosing the right filter actions
Building a comprehensive safety strategy
Balancing automation with human moderation

Understanding Content Filtering

How Filtering Works

InterChat's content filtering system operates in real-time:

Message Analysis - Every message is checked against your filter patterns
Pattern Matching - Text is compared using various matching techniques
Action Execution - Violations trigger automated responses
Logging - All filter actions are recorded for review

Filter Actions Explained

Prevents message from appearing

When to use:

Obvious inappropriate content
Spam patterns
Prohibited links or content
Clear rule violations

What happens:

Message is stopped before reaching other servers
Original sender sees their message normally
Other servers never see the blocked content
Action is logged for moderator review

Best for: High-confidence violations that should never appear

Automatically removes user from hub (10 minutes)

When to use:

Severe violations (hate speech, harassment)
Spam bots or malicious accounts
Content that requires immediate removal
Patterns indicating bad faith participation

What happens:

User is immediately blacklisted for 10 minutes
All their recent messages are flagged for review
They cannot participate until blacklist expires
Moderators are notified of the action

Best for: Serious violations requiring immediate action

Notifies moderators for human review

When to use:

Borderline content needing context
New or experimental filter patterns
Content that might have false positives
Situations requiring human judgment

What happens:

Message appears normally in all servers
Alert sent to moderators with violation details
Moderators can take action if needed
Helps refine filter patterns over time

Best for: Content that needs human evaluation

Setting Up Content Filtering

Access Filter Configuration

/hub config anti-swear hub:YourHubName

This opens the content filtering management interface.

Create Your First Filter

Click "Add New Filter"
Enter your pattern (see pattern guide below)
Choose the appropriate action
Add a description for your team
Save the filter

Test Your Filters

Start with "Send Alert" action for new patterns
Test with trusted community members
Monitor alerts for false positives
Adjust patterns based on results

Refine and Optimize

Review filter logs regularly
Update patterns based on new violations
Remove or modify filters causing false positives
Share effective patterns with your moderation team

Pattern Matching Guide

Pattern Types

Pattern: word

Matches: Only the exact word "word"

Examples:

✅ "word" (matches)
❌ "words" (doesn't match)
❌ "keyword" (doesn't match)
❌ "sword" (doesn't match)

Best for:

Specific prohibited terms
Exact phrases or commands
Brand names or specific references
When precision is critical

Use case: Blocking specific slurs or exact spam phrases

Pattern: word*

Matches: Any word starting with "word"

Examples:

✅ "word" (matches)
✅ "words" (matches)
✅ "wordsmith" (matches)
❌ "keyword" (doesn't match)
❌ "sword" (doesn't match)

Best for:

Word variations and plurals
Terms with common suffixes
Catching creative spelling attempts
Broad category filtering

Use case: Blocking "spam*" to catch "spam", "spammer", "spamming"

Pattern: *word

Matches: Any word ending with "word"

Examples:

✅ "word" (matches)
✅ "keyword" (matches)
✅ "password" (matches)
❌ "words" (doesn't match)
❌ "wordsmith" (doesn't match)

Best for:

Common word endings
Suffix-based violations
Catching variations of prohibited terms
Technical term filtering

Use case: Blocking "*bot" to catch "spambot", "autobot", etc.

Pattern: *word*

Matches: Any text containing "word"

Examples:

✅ "word" (matches)
✅ "words" (matches)
✅ "keyword" (matches)
✅ "sword" (matches)
✅ "wordsmith" (matches)

Best for:

Broad content filtering
Catching creative spelling
General topic restrictions
When context doesn't matter

Use case: Blocking "discord.gg" to catch all Discord invites

⚠️ Warning: Can cause many false positives

Advanced Pattern Strategies

Layered Filtering Approach:

Exact matches for known violations
Prefix/suffix matches for variations
Contains matches for broad categories
Alert actions for borderline content

Example Filter Set for Spam Prevention:

Pattern: "discord.gg" | Action: Block | Description: Discord invite links
Pattern: "bit.ly*" | Action: Send Alert | Description: Shortened URLs for review
Pattern: "*free nitro*" | Action: Blacklist | Description: Common scam phrase
Pattern: "join my server" | Action: Send Alert | Description: Potential server advertising

Building Effective Filter Lists

Common Filter Categories

Profanity and Hate Speech:

Approach:

Start with exact matches for clear violations
Use prefix/suffix for variations
Consider context and community standards
Regular review and updates needed

Example Patterns:

[slur] | Block | Exact hate speech term
[profanity]* | Block | Profanity and variations
*[offensive-term]* | Send Alert | Borderline content

Best Practices:

Research community-specific terms
Consider cultural and regional differences
Balance strictness with false positives
Have clear appeal processes

Spam and Scam Prevention:

Common Spam Patterns:

*free nitro* | Blacklist | Nitro scams
*click here* | Send Alert | Suspicious links
discord.gg | Block | Invite links
*dm me* | Send Alert | Potential spam
*check my bio* | Send Alert | Profile spam

Scam Indicators:

Promises of free premium services
Urgent action required language
External link requests
Too-good-to-be-true offers

Strategy:

Block obvious scams immediately
Alert on suspicious patterns
Monitor for new scam techniques

NSFW Content Filtering:

Technical Approach:

Enable built-in NSFW detection
Add text-based NSFW filters
Consider community standards
Clear policies on borderline content

Example Filters:

*nsfw* | Block | NSFW references
*18+* | Send Alert | Age-restricted content
[explicit-terms] | Block | Sexual content

Considerations:

Community age demographics
Professional vs. casual environments
Cultural sensitivity
Clear content policies

Server Advertising Prevention:

Advertising Patterns:

discord.gg | Block | Discord invites
*join my server* | Send Alert | Direct advertising
*new server* | Send Alert | Server promotion
*looking for members* | Send Alert | Recruitment

Balanced Approach: