Text Anti-Slop Skill for Technical Writing

When to Use This Skill

Use text-anti-slop when:

• ✓ Writing technical documentation, README files, or guides
• ✓ Reviewing AI-generated documentation before publishing
• ✓ Cleaning up marketing language in technical content
• ✓ Editing academic or research writing
• ✓ Preparing content for professional publication
• ✓ Teaching writing standards to teams

Do NOT use when:

• Writing creative fiction (different standards)
• Legal documents (require specific phrasings)
• Marketing copy (intentionally uses persuasive language)
• Following specific style guides (AP, Chicago, etc.)

For the 24-pattern Wikipedia humanizer checklist, use the separate humanizer skill.

Quick Example

Before (AI Slop):

In this document, we will delve into the complexities of API design. It's important to note that in today's fast-paced world, leveraging best practices is crucial for success. We'll navigate through various approaches, ultimately providing you with actionable insights to empower your development workflow.

After (Anti-Slop):

This guide covers API design principles. We explain three approaches: RESTful, GraphQL, and gRPC. Each section includes code examples and trade-offs.

What changed:

• ✓ Removed meta-commentary ("In this document, we will...")
• ✓ Eliminated filler ("delve into", "navigate through")
• ✓ Replaced buzzwords ("leverage", "empower")
• ✓ Removed obvious transitions ("ultimately", "it's important to note")
• ✓ Led with the point directly

When to Use What

Core Workflow

3-Step Quality Check

1. Remove meta-commentary

   markdownCopy

   # Bad
   In this section, we will discuss error handling.
   Let's take a closer look at the implementation.
   
   # Good
   Error handling uses try/catch blocks.
   The implementation checks for three error types:
   

2. Replace filler and buzzwords

   markdownCopy

   # Bad
   In order to leverage the power of our framework, users need to
   utilize the config file to enable the feature.
   
   # Good
   To use the framework, set `enable_feature: true` in `config.yml`.
   

3. Lead with the point

   markdownCopy

   # Bad
   It's important to note that before you proceed with installation,
   you should be aware that Python 3.8+ is required.
   
   # Good
   Requires Python 3.8 or later.
   

Quick Reference Checklist

Before publishing technical writing, verify:

• No meta-commentary ("In this guide...", "Let's explore...")
• No high-risk transitions ("delve into", "navigate complexities")
• No corporate buzzwords ("leverage", "synergistic", "paradigm shift")
• No wordy constructions ("in order to" → "to")
• Paragraphs lead with the point, not preambles
• Concrete examples, not abstract statements
• Varied sentence structure and length
• Active voice where possible
• Specific terms, not generic ("data", "things", "items")
• Direct statements, not excessive hedging

Common Workflows

Workflow 1: Clean Up README or Documentation

Context: AI generated a README with generic patterns.

Steps:

1. Run detection script

   bashCopy

   python toolkit/scripts/detect_slop.py README.md --verbose
   

2. Remove meta-commentary

   markdownCopy

   # Before
   ## Introduction
   In this project, we aim to provide developers with a comprehensive
   solution for managing configuration files. Let me walk you through
   the key features that make this tool essential.
   
   # After
   ## Features
   - Hot reload configuration without restart
   - Environment-specific configs
   - Schema validation with JSON Schema
   

3. Replace buzzwords

   markdownCopy

   # Before
   Our solution leverages cutting-edge technology to empower developers
   and unlock the full potential of configuration management.
   
   # After
   Validates and reloads config files automatically.
   

4. Simplify wordy phrases

   markdownCopy

   # Before
   Due to the fact that the system has the ability to detect changes,
   it will automatically reload in order to apply the new settings.
   
   # After
   The system detects changes and reloads automatically.
   

5. Run automated cleanup

   bashCopy

   python toolkit/scripts/clean_slop.py README.md --save
   

Expected outcome: Clear, direct documentation with score <30

Workflow 2: Review Technical Blog Post

Context: Cleaning up a technical blog post before publication.

Steps:

1. Check opening paragraphs

   markdownCopy

   # Bad
   In today's fast-paced world of software development, it's become
   increasingly important to understand the complexities of async programming.
   In this article, I will delve into the intricacies of async/await.
   
   # Good
   Async/await lets JavaScript handle concurrent operations without blocking.
   This post shows three common patterns and their trade-offs.
   

2. Remove transition overload

   markdownCopy

   # Bad
   Furthermore, it's worth noting that moreover, the implementation
   ultimately provides developers with enhanced capabilities.
   
   # Good
   The implementation supports retries, timeouts, and cancellation.
   

3. Replace vague terms with specifics

   markdownCopy

   # Bad
   The data shows interesting results. Various metrics indicate
   significant improvements. Things are looking positive.
   
   # Good
   Response time dropped from 200ms to 45ms. Memory usage decreased 30%.
   

4. Check closing

   markdownCopy

   # Bad
   In conclusion, as we've explored throughout this article, it's clear
   that async programming plays a crucial role in modern development.
   
   # Good
   Use async/await for I/O operations. Avoid it for CPU-intensive work.
   

Expected outcome: Professional blog post that teaches, not pads

Workflow 3: Clean Up API Documentation

Context: Auto-generated API docs need human touch.

Steps:

1. Fix function descriptions

   markdownCopy

   # Bad
   ### `processData(data)`
   This function processes the data. It takes data as input and returns
   the processed result. This is a critical function for data processing.
   
   # Good
   ### `processData(data)`
   Validates and normalizes customer records.
   
   **Parameters:**
   - `data` (Array): Raw customer records from CSV import
   
   **Returns:** Array of validated records with normalized phone numbers
   

2. Add concrete examples

   markdownCopy

   # Bad
   **Example:**
   ```js
   processData(data)
   

   Good

   Example:

   jsCopy

   const raw = [{name: "Alice", phone: "555.1234"}]
   const clean = processData(raw)
   // => [{name: "Alice", phone: "555-1234"}]
   

   Copy

3. Remove obvious headers

   markdownCopy

   # Bad
   ## Overview
   This section provides an overview of the API.
   
   ## Getting Started
   To get started with the API, follow these steps.
   
   # Good
   ## Authentication
   Include your API key in the `Authorization` header:
   

Expected outcome: Docs that help developers, not slow them down

Mandatory Rules Summary

1. No Meta-Commentary

Delete stage directions about what the document will do

Bad: "In this section..." "Let's explore..." "As we'll see..." Good: Just say the thing.

2. Lead with the Point

First sentence states the main idea

Bad: Preamble → context → point Good: Point → details

3. Specific Not Generic

Use concrete terms

Bad: "data", "things", "items", "various" Good: "customer records", "API endpoints", "config files"

4. Simple Not Wordy

Choose shorter alternatives

• "in order to" → "to"
• "due to the fact that" → "because"
• "has the ability to" → "can"

5. Direct Not Hedged

Be confident or be specific about uncertainty

Bad: "may possibly", "could potentially", "it appears that" Good: "does X" or "unclear whether X"

Resources & Advanced Topics

Reference Files

• reference/transitions.md - Overused transition phrases to avoid
• reference/buzzwords.md - Corporate jargon and replacements
• reference/filler.md - Wordy constructions and alternatives
• reference/meta-commentary.md - Self-referential patterns to delete
• reference/structure.md - Document organization principles

Related Skills

• humanizer - Wikipedia's 24-pattern checklist for AI writing
• r/anti-slop - For R package documentation (roxygen, vignettes)
• python/anti-slop - For Python docstrings
• quarto/anti-slop - For reproducible research documents

Tools

• python toolkit/scripts/detect_slop.py - Detect text patterns (score 0-100)
• python toolkit/scripts/clean_slop.py - Automated cleanup with backup

Scoring Guide

Detection script scores (0-100, higher is worse):

Context Awareness

Not all patterns are always slop. Consider:

• Audience: Academic writing may need more hedging
• Purpose: Legal docs need specific phrasing; marketing needs energy
• Length: Longer pieces need more transitions
• Domain: Technical docs have different norms than creative writing

The issue is overuse and unconscious repetition, not pattern existence.

Integration with Humanizer Skill

This skill focuses on structural clarity (remove transitions, buzzwords, filler).

The humanizer skill adds personality and voice (Wikipedia's 24 patterns).

Use both for complete coverage: