Slop Blocker par mx92

WHAT IT DOES
• Scores every reply using multiple signals: suspicious username patterns, missing profile pictures, link spam, and engagement-farming phrases
• Optionally routes ambiguous replies to an AI judge via your own API key (BYOK). No data ever touches our servers
• Soft-hides replies with a "show anyway" option. No reply is permanently deleted
• Shows a reason for every hidden reply so you know exactly why it was filtered

WHY IT'S DIFFERENT
• 100% open source (AGPL v3). Review every line at github.com/mx-92x/slop-blocker
• Local-first: no account, no backend, no telemetry. Everything runs in your browser
• Fully tunable: adjust scoring thresholds with sliders in the popup
• BYOK (Bring Your Own Key): optionally connect OpenAI, Anthropic, DeepSeek, Groq, Mistral, or any OpenAI-compatible endpoint

PRIVACY
All filtering happens on your device. If you use BYOK, your text goes directly to your chosen provider. Slop Blocker never sees it. API keys are encrypted with AES-GCM before storage.