Twitter Hate Blocker

A Chrome extension that scans replies on Twitter/X, detects hateful content using AI, and automatically blocks or reports offending accounts. Zero configuration required.

Features

• AI-powered detection with server-side classification (no API keys needed)
• Three detection modes:
  • Hate Speech — Flags direct hatred, slurs, and harassment targeting you
  • Cult Praise — Flags abnormally sycophantic, cult-like devotion
  • Block All — Blocks every account on the page (useful for hashtag cleanup)
• Three action modes: Block, Report, or Both
• Auto-scroll to find more replies beyond the initial page load
• Live results panel showing flagged accounts with confidence scores and reasons
• Dry run mode to preview what would be blocked without taking action
• Bilingual UI — English and Persian (فارسی)
• Configurable scan limit (25, 50, or 100 replies)
• Daily usage tracking to stay within platform limits

Installation

Chrome Web Store

Coming soon.

Manual Installation

1. Clone this repo
2. Install dependencies:

   pnpm install

3. Build the extension:

   pnpm build

4. Open chrome://extensions and enable "Developer mode"
5. Click "Load unpacked" and select the .output/chrome-mv3 folder

How It Works

1. Navigate to a tweet with replies on Twitter/X
2. Click the scan button that appears on the page
3. The extension collects replies and sends the text to a Cloudflare Worker backend
4. The backend classifies each reply using AI (Gemma 3 12B via OpenRouter)
5. Flagged accounts are automatically blocked, reported, or both based on your settings

Content Script → Background Worker → Cloudflare Worker → OpenRouter API
     ↓                    ↓                  ↓
  Twitter DOM         Proxies request    Rate limits, classifies
                      with clientId      Returns: {isMatch, confidence, reason}

Development

# Dev with hot reload
pnpm dev

# Run tests
pnpm test

# Build for production
pnpm build

# Package as zip
pnpm zip

Tech Stack

• Extension: WXT + React + TypeScript (Manifest V3)
• Backend: Cloudflare Worker + D1 (SQLite)
• AI Model: Gemma 3 12B via OpenRouter
• Testing: Vitest

Privacy

Tweet text is sent to a secure backend for AI classification only. No data is stored beyond rate limiting. No login or account required. See the full Privacy Policy.

Support This Project

If you find this project useful, consider supporting its development:

• Star this repo
• Report bugs and submit PRs
• Sponsor on GitHub or Buy me a coffee

License

MIT