De-Slop

A Chrome extension that detects and removes AI-generated "slop" from web pages. 600+ patterns across 11 languages, 8 analysis tools, 3 themes, zero API calls. Runs entirely in your browser.

---

Features

Core Detection

• Three-tier detection system with weighted pattern matching (600+ patterns)
• Multi-language support - 11 languages: English, Spanish, French, German, Italian, Portuguese, Swedish, Polish, Japanese, Korean, Chinese
• Real-time scanning of dynamically loaded content
• Badge counter showing blocked slop on current page
• Configurable sensitivity (1-5 slider)
• Detection Only Mode - highlight slop instead of removing it, with hover tooltips
• Custom pattern editor - add your own slop patterns with custom weights
• Three themes - Dark, Light, and Midnight (OLED true black), synced across all pages

Analysis Tools

• Slop Checker - paste text and see what gets flagged with real-time highlighting and suggestions
• Humanize Score - 8-metric writing quality analysis with authenticity scoring
• Batch Checker - score multiple text blocks at once with CSV export
• Text Rewriter - inline fix suggestions for detected slop phrases
• Text Comparator - side-by-side draft comparison with score deltas
• URL Analyzer - fetch and score any webpage for slop content
• Slop Machine - gamified slot machine for learning better word choices
• Pattern Tester - validate detection patterns across all 11 languages

LinkedIn Fixer

When on LinkedIn.com, additional features activate:

• Block auto-play videos
• Block ad tracking
• Darker mode - true black LinkedIn theme

Advanced Features

• Emoji slop detection - flag emoji spam and engagement bait (166 patterns)
• Stop-word filtering - detect "I'm excited to announce" style posts
• Em dash detector - AI loves em dashes
• Political content filter - optional political content detection
• YouTube filtering - video titles, descriptions, Shorts, and clickbait detection
• Improvement suggestions - better alternatives for sloppy phrases
• Export/import custom pattern sets

Installation

From Chrome Web Store

Coming soon

Development Mode

1. Clone this repository
2. Open Chrome and go to chrome://extensions/
3. Enable "Developer mode" (top-right)
4. Click "Load unpacked"
5. Select the src directory

Detection System

De-Slop uses a three-tier aggression system:

Tier 1: AI-Specific Indicators (3 points each)

Signature AI phrases and patterns:

• "delve into", "navigate the landscape", "tapestry of", "realm of possibilities"
• Em dash overuse, excessive transitions
• "In today's fast-paced world...", "As we continue to evolve..."
• Stop words: "I'm excited to announce", "Thrilled to share"
• Listicle patterns: "Top 10 ways to...", "Ultimate guide to..."

Active at all sensitivity levels

Tier 2: Corporate Buzzwords (2 points each)

LinkedIn's greatest hits:

• "synergy", "leverage", "circle back", "low-hanging fruit"
• "take this offline", "touch base", "thought leadership"
• "AI-powered", "blockchain-enabled", "digital transformation"

Active at sensitivity 3+

Tier 3: Marketing Spam (1 point each)

The widest net:

• "free", "guaranteed", "limited time offer", "buy now"
• "amazing", "incredible", "revolutionary", "mind-blowing"
• "basically", "essentially", "actually"

Active at sensitivity 4+

Scoring Thresholds

Sensitivity	Threshold	Description
1	15 points	Very conservative - obvious AI only
2	12 points	Conservative
3	9 points	Balanced (default) - AI + corporate
4	6 points	Aggressive - includes marketing
5	4 points	Nuclear - maximum detection

Usage

Popup (Dark & Light themes)

Toggle on/off, adjust sensitivity, access all tools. Cycle themes with the sun icon in the header.

Analysis Tools

Humanize Score -- 8-metric writing quality analysis

Batch Checker -- score multiple text blocks at once

Text Rewriter -- side-by-side input with live fix suggestions

Pattern Tester

Validate detection patterns across all 11 supported languages.

Settings & Theme Selector

Configure patterns, sensitivity, language, and choose your theme (Dark, Light, Midnight).

About

LinkedIn Fixer

When on LinkedIn, additional toggles appear:

• Block auto-play videos
• Block ad tracking
• Enable darker mode (true black theme)

How It Works

1. Scans page content using platform-specific selectors
2. Assigns weighted "slop score" based on pattern matches
3. Removes elements exceeding threshold
4. Updates badge counter
5. Monitors for dynamic content (infinite scroll, AJAX)

Platform Support

• LinkedIn: .feed-shared-update-v2 (individual posts)
• Twitter/X: [data-testid="tweet"]
• Medium: article, .postArticle
• YouTube: Videos, Shorts, comments, clickbait detection
• All sites: Generic article/post/card selectors

Repository Structure

DeSlop/
├── src/
│   ├── manifest.json        # Extension configuration (Manifest V3)
│   ├── theme.css            # CSS variable definitions (dark/light/midnight)
│   ├── theme-loader.js      # Sync theme loader (prevents flash)
│   ├── background.js        # Service worker
│   ├── content.js           # Main detection engine
│   ├── content.css          # Styles for hiding/highlighting slop
│   ├── scoring-engine.js    # Shared scoring module
│   ├── i18n-helper.js       # Internationalization helper
│   ├── languages.js         # Language definitions
│   ├── patterns/            # Externalized pattern system
│   │   ├── registry.js      # Pattern registry & language resolution
│   │   ├── en.js            # English patterns (600+)
│   │   ├── es.js, fr.js ... # 10 additional languages
│   ├── _locales/            # Chrome i18n message files (11 locales)
│   ├── popup.*              # Main sidebar interface
│   ├── checker.*            # Interactive slop checker
│   ├── humanize.*           # Writing quality analyzer
│   ├── batch.*              # Batch text analysis
│   ├── rewriter.*           # Inline fix suggestions
│   ├── compare.*            # Side-by-side comparison
│   ├── url-analyzer.*       # URL content scoring
│   ├── slop-machine.*       # Gamified learning tool
│   ├── test.*               # Multi-language pattern tester
│   ├── settings.*           # Pattern customization UI
│   ├── about.*              # Project info and attribution
│   ├── linkedin-fixer.js    # LinkedIn-specific enhancements
│   ├── rules.json           # Ad blocking rules
│   └── icons/               # Extension icons
├── assets/                  # Screenshots and marketing images
├── privacy.html             # Privacy policy
├── LICENSE                  # GPL-3.0 license
└── README.md

Privacy

• Runs entirely locally - no data sent to servers
• Stores settings in Chrome sync storage
• No analytics or tracking
• No external API calls
• Open source - inspect the code

Permissions

• storage - Save your settings
• activeTab - Access current page content
• scripting - Dynamic language pattern injection
• sidePanel - Sidebar interface
• declarativeNetRequest - Block LinkedIn ads (optional)
• <all_urls> - Scan any page for slop

Known Limitations

• Pattern-based detection may have false positives
• Very short content (<100 chars) skipped to reduce false positives
• Platform selectors may need updates when sites change
• If your entire LinkedIn feed disappears - it's all slop! Lower sensitivity.

Version History

v3.1.0 - Theme System & About Page

• Three-theme system: Dark (default), Light, and Midnight (OLED true black)
• CSS variable architecture across all 11 UI pages (~35 design tokens)
• Theme cycle button in popup header, theme selector cards in settings
• Sync theme loader prevents flash of wrong theme on page load
• Theme persists across pages and browser restarts via chrome.storage.sync
• About page with project info, feature summary, and privacy statement
• Improved button contrast for WCAG accessibility
• Removed stale release files

v3.0.0 - Multi-language & Tools Suite

• 11-language pattern system (en, es, fr, de, it, pt, sv, pl, ja, ko, zh)
• Full i18n with localized UI
• 6 new analysis tools (humanize, batch, rewriter, compare, url-analyzer, pattern tester)
• Externalized pattern architecture with shared scoring engine
• Expanded settings with pattern management overhaul

v2.7.2 - Political Content Filter & YouTube

• Political content filter
• YouTube video/Shorts/comment filtering
• Clickbait pattern detection
• Improvement suggestions

v2.4.4 - Detection Only Mode

• Highlight-only mode (non-destructive detection)
• Emoji detection expanded to 166 patterns
• Simplified UI with sensitivity-driven tier control

v2.0.0 - LinkedIn Fixer & Dark Theme

• Complete UI redesign with dark theme
• Pattern customization system with import/export
• LinkedIn video blocking, ad blocking, darker mode

v1.0.0 - Initial release

• Three-tier detection system
• Basic popup controls

License

GNU General Public License v3.0 (GPL-3.0)

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

See LICENSE file for full license text.

Credits

Made by Kief Studio | Developed by HxHippy