Claude Slop Detector

A Claude Code skill for detecting AI-generated writing patterns ("slop") in content.

What is "Slop"?

AI slop is content that sounds generically AI-generated — the kind of writing that could be sent to anyone in your industry without modification. It's characterized by:

• Overused phrases like "delve into," "game-changer," and "it's worth noting"
• Uniform sentence lengths (every sentence 10-15 words)
• Staccato fragment spam (short. punchy. sentences. everywhere.)
• Over-balanced sections with diplomatic hedging
• Manufactured enthusiasm and fake personality

This skill helps you identify these patterns and fix them.

Installation

Option 1: Skills CLI

npx skills add aplaceforallmystuff/claude-slop-detector

Option 2: Copy to your Claude Code skills directory

# Clone the repository
git clone https://github.com/aplaceforallmystuff/claude-slop-detector.git

# Copy to your Claude Code skills directory
cp -r claude-slop-detector/skills/slop-detector ~/.claude/skills/

Option 3: Clone directly to skills directory

git clone https://github.com/aplaceforallmystuff/claude-slop-detector.git ~/.claude/skills/slop-detector

Option 4: Manual installation

1. Create the directory: mkdir -p ~/.claude/skills/slop-detector
2. Download SKILL.md to that directory

Usage

Once installed, the skill activates when you ask Claude Code to:

• "Check this draft for AI slop"
• "Audit this content for AI patterns"
• "Run slop detection on this file"
• "Is this writing too AI-sounding?"

Example

> Check this draft for slop: "In today's digital landscape, it's worth noting
  that AI has become a game-changer for content creation. Let's delve into
  how you can leverage these cutting-edge solutions to unlock your potential."

Output:

## Slop Detection Report

**30-Second Test:** FAIL - could be sent to anyone
**Overall Slop Score:** 21 - High Risk

### Tier 1 Findings (Remove Immediately)
- "delve into" - Replace with: examine, explore, look at
- "game-changer" - Replace with: specific impact description
- "it's worth noting" - Remove entirely
- "cutting-edge solutions" - Replace with: specific tools/methods
- "unlock your potential" - Replace with: specific outcome
...

How It Works

The skill uses a tiered detection system:

Tier 1: Almost Always AI (Remove Immediately)

Phrases so strongly associated with AI that their presence alone suggests unedited output:

• "delve into" / "delve"
• "game-changing" / "revolutionary"
• "it's worth noting that"
• "moreover" / "furthermore"
• "unlock your potential"
• "cutting-edge solutions"

Research shows "delve" usage increased 10.45x post-ChatGPT (Finnish study, 56,878 essays).

Tier 2: Suspicious (Check Context)

Problematic when overused or clustered:

• "here's the thing" (repeated)
• "at the end of the day"
• "comprehensive and thorough"
• "let's dive in"

Tier 3: Watch for Patterns

Fine individually, problematic in clusters:

• "however" starting every paragraph
• "firstly," "secondly," "thirdly"
• "robust," "seamless," "scalable" together

Structural Tells

Beyond phrases, the skill detects:

• Sentence uniformity - all sentences same length
• Staccato fragment spam - "Short. Punchy. Sentences."
• Over-balanced sections - diplomatic hedging everywhere
• List addiction - bullets where prose works better
• Manufactured personality - fake developer voice

Slop Score

The skill calculates a numerical score:

The 30-Second Test

The fastest way to identify AI slop:

"Could this exact content be sent to anyone in my industry?"

If yes, it's slop. Specificity separates genuine insight from generic templates.

License

MIT License - see LICENSE

Acknowledgments

Detection patterns informed by:

• Finnish study on post-ChatGPT writing changes (56,878 essays)
• Georgia Tech analysis of 168.3M academic articles
• Biomedical literature analysis of AI marker co-occurrence

Related Skills

Part of the aplaceforallmystuff skills collection:

• the-antislop — More comprehensive detection + editor mode (35+ patterns)
• voice-analyzer — Create a voice profile from your writing samples
• voice-editor — Edit content to match your voice profile

The goal isn't perfect avoidance of any single phrase — it's writing that sounds like a specific human with specific opinions, not a very polite committee trying not to offend anyone.