The Context Problem

Traditional automod filters are dumb. They look for specific "bad words."

• User says: "I hate ice cream." -> Banned (contains "hate").
• User says: "I am going to un-alive you." -> Ignored (no bad words).
  This creates false positives and lets toxic behavior slide. You need an automod that understands Context.

Enter AI Sentiment Analysis

SkynetBot (Tier 1+) integrates with Google Cloud Natural Language API to analyze the emotion behind a message, not just the keywords.

How it Works

1. Analysis: Every message is scored from -1.0 (Very Negative) to +1.0 (Very Positive).
2. Thresholds: You set the limit.
   • Strict: Block anything below -0.5.
   • Lenient: Only block extreme toxicity (-0.9).
3. Categories: The AI detects:
   • Toxicity
   • Insults
   • Threats
   • Identity Attacks (Racism/Homophobia)

Configuration

Enable it in Dashboard > Moderation > Sentiment Filter.
You can choose to:

• Warn: "Hey, cool it down."
• Delete: Silently remove the message.
• Mute: Timeout the user for 5 minutes.
  Stop playing whack-a-mole with word lists. Let AI handle the toxicity.