anti-ai-writing

Transform AI-assisted drafts into authentic, human-sounding content. This skill provides patterns to detect and eliminate AI tells, frameworks for natural writing, and techniques for creating prose that reads as genuinely human. Use when reviewing any AI-generated content or when writing content that must not appear AI-assisted.

8 stars

Best use case

anti-ai-writing is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Transform AI-assisted drafts into authentic, human-sounding content. This skill provides patterns to detect and eliminate AI tells, frameworks for natural writing, and techniques for creating prose that reads as genuinely human. Use when reviewing any AI-generated content or when writing content that must not appear AI-assisted.

Teams using anti-ai-writing should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/anti-ai-writing/SKILL.md --create-dirs "https://raw.githubusercontent.com/cdeistopened/skill-stack/main/.claude/skills/anti-ai-writing/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/anti-ai-writing/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How anti-ai-writing Compares

Feature / Agentanti-ai-writingStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Transform AI-assisted drafts into authentic, human-sounding content. This skill provides patterns to detect and eliminate AI tells, frameworks for natural writing, and techniques for creating prose that reads as genuinely human. Use when reviewing any AI-generated content or when writing content that must not appear AI-assisted.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

# Anti-AI Writing Engine

Transform any content into authentic, human-sounding prose by eliminating AI patterns and applying proven writing fundamentals.

## Purpose

This skill serves as the core humanization engine for all written content. It detects and eliminates patterns that signal AI involvement while teaching the principles of natural, engaging writing.

**Core Philosophy:** The best writing is invisible. Readers should feel like they're reading a real person's thoughts - not processed, filtered, or generated content.

## When to Use This Skill

- Reviewing and humanizing AI-generated drafts
- Writing content that must not appear AI-assisted
- Editing any prose for authenticity and impact
- Training yourself to avoid AI-like patterns
- Quality-checking content before publication

**Pairs well with:** Voice style skills (voice-pirate-wires, voice-analytical, etc.) for writing in a specific author's style.

---

## Universal Writing Fundamentals

Master these principles first, then layer on specific style choices.

### 1. Energy Transfer

**The best writing is a transfer of energy from writer to reader.**

- Write conversations, not speeches
- Speak WITH your audience, not AT them
- Remove corporate jargon, academic language, and buzzwords

**Example:**
- ❌ "The implementation of our innovative frameworks has demonstrated significant positive outcomes in terms of engagement metrics."
- ✅ "Our new approach works. People are paying attention."

The second version transfers energy. Readers feel conviction, not formality.

### 2. Conciseness is King

**Every word must earn its place.**

**Process:**
1. Write your first draft freely
2. Read it aloud
3. Delete anything that:
   - Repeats an idea already stated
   - Softens your main point
   - Sounds formal instead of conversational
   - Takes more words than necessary

**Example:**
- ❌ "Our approach incorporates multiple methodologies designed to accommodate diverse requirements and development patterns."
- ✅ "We adapt to how you work."

### 3. The SUCKS Framework

Before writing anything, answer these questions:

**S - Specific**
- Write for ONE person, not a crowd
- Know their struggles, goals, language
- If you write for everyone, you write for no one

**U - Unique & Useful**
- Does this change how your reader thinks, feels, or acts?
- Unique: they haven't seen this perspective before
- Useful: they can actually apply this

**C - Clear, Curious, Conversational**
- Clear: Main idea is unmissable
- Curious: Questions that make readers think
- Conversational: Reads like talking to a friend

**K - Kept Simple & Structured**
- Simple ideas, simple words
- Clear beginning, middle, end

**S - Sticky**
- Memorable phrases they'll repeat
- Ideas that lodge in memory

**Application checklist:**
- [ ] Who is my ONE reader?
- [ ] What will they think, feel, or do differently?
- [ ] Could a 12-year-old understand the main point?
- [ ] Does this flow like a conversation?

### 4. Be Undeniable

**Transform vague claims into concrete specifics.**

- ❌ "How to boost your results by at least 87% in just one month."
- ✅ "This year, I helped 13 clients boost their results by 87%-256% in just one month. Here's how:"

**Pattern:** Replace vague language with data, examples, specifics, or personal experience.

### 5. Craft Sticky Sentences

Use these techniques to create memorable statements:

**Alliteration** - Same starting sounds
- "Specificity is the secret"
- "The best jobs are neither decreed nor degreed"

**Symmetry** - Parallel structure
- "Read for awareness. Write for understanding."
- "It's not 10,000 hours. It's 10,000 iterations."

**Contrast** - Opposing ideas
- "To be everywhere is to be nowhere."
- "Be clear, not clever. Concise, not complex."

**Rhyme** - Similar ending sounds
- "Tell a story or lose your glory"

**Rhythm** - Pleasing cadence
- "When you can't wait to share it, they can't help but read it"

**For maximum stickiness, combine techniques:**
- "Anger prepares us to fight. Fear prepares us to flee." (Symmetry + Alliteration)

See `references/sticky-sentences.md` for expanded examples.

---

## Forbidden Patterns (AI Tells)

These patterns consistently signal AI involvement. Eliminate them aggressively.

### The Correlative Construction (Most Common AI Tell)

This is the #1 pattern to avoid:

- ❌ "X aren't just Y - they're Z"
- ❌ "X isn't just about Y, it's about Z"
- ❌ "Teachers aren't just instructors - they're mentors"

**Why it's bad:** This is the default structure AI models produce. Every AI writing sample uses this.

**Better alternatives:**
- ✅ "Teachers mentor students"
- ✅ "X accomplishes both Y and Z"
- ✅ State the point directly

### Forbidden Constructions

- ❌ "It's not about X, it's about Y"
- ❌ "The truth is..." / "The reality is..."
- ❌ "Let that sink in"
- ❌ "Now more than ever..."

**Better:** State your point directly without setup.

### Forbidden Rhetorical Flourishes

- ❌ "The best part? ..."
- ❌ "The secret? ..."
- ❌ "Sounds impossible? It's not."
- ❌ "What if I told you..."
- ❌ "Here's the thing..."
- ❌ "Let's be honest..."
- ❌ "At the end of the day..."

### Forbidden Staccato Phrasing

Do not use faux-dramatic sentence fragments:
- ❌ "No fluff. No filler. Just results."
- ❌ "Big ideas. Small words. Maximum impact."
- ❌ "Simple. Clear. Effective."
- ❌ "Stop guessing. Start winning."

### Forbidden Openers

- ❌ "In the ever-evolving world of..."
- ❌ "In today's fast-paced..."
- ❌ "Gone are the days when..."
- ❌ "This underscores..."
- ❌ "In an era where..."

### Formatting Restrictions

- Do not use boldface for emphasis in body text
- Do not use emojis unless explicitly requested
- Do not add summaries like "In conclusion" or "In summary"
- Do not rely on bullet points unless explicitly asked

### Dash Usage

Use regular hyphens with spaces for parenthetical breaks - like this - not em dashes (—) or en dashes (–).

- ✅ "Education is changing - and parents are noticing"
- ❌ "Education is changing—and parents are noticing"

---

## Anti-AI Pattern Detection

Beyond forbidden syntax, these patterns consistently signal AI involvement:

### 1. Overuse of "Just"

**Problem:** AI uses "just" constantly as a softener.

- ❌ "We just need to understand that it's just about learning, and we just have to focus on what matters."
- ✅ "It's about learning. Focus on what matters."

**When "just" is acceptable:** "just one question," "just in case," "just started"

**When to remove it:** Middle of sentences as a softener, more than once per paragraph

### 2. Overuse of "Actually"

**Problem:** AI uses "actually" to sound like it's adding nuance.

- ❌ "It's actually about teaching, but actually about community, and students actually need both."
- ✅ "It's about teaching. It's also about community. Students need both."

**When acceptable:** Correcting a previous misunderstanding

**Limit:** Once per 500 words maximum

### 3. Hedging Language

AI loves hedge words because they make uncertain claims safer. But they make writing weak.

**Common hedges to eliminate:**
- ❌ "This might help you" → ✅ "This will help you"
- ❌ "Some people say..." → ✅ "I've found that..."
- ❌ "It could be argued that..." → ✅ "It's true that..."
- ❌ "Perhaps we should consider..." → ✅ "We should..."
- ❌ "One could say..." → ✅ "This is true because..."

**Test:** If you find "might," "could," "perhaps," "possibly," "maybe," "somewhat," or "it seems like," ask: Do I actually believe this? If yes, remove the hedge.

### 4. Passive Voice

**Problem:** AI defaults to passive voice because it's safer.

- ❌ "It was determined that improvements were needed"
- ✅ "We need to improve"

- ❌ "It has been shown that users benefit from..."
- ✅ "Users benefit from..."

**Test:** If you can't identify who's doing the action, it's passive. Rewrite with a subject.

### 5. Corporate Jargon

These signal over-processing:

- ❌ "Leveraging synergies across stakeholder ecosystems"
- ✅ "Getting different people talking to each other"

- ❌ "Optimize engagement metrics"
- ✅ "Get people paying attention"

**Test:** Would you say this to a friend? If not, rewrite it plainly.

### 6. Vague Language

AI tends toward vagueness because it's safer. Vague writing is weak writing.

- ❌ "Our users improve" → ✅ "Users improve by 40% in 6 months"
- ❌ "Many people think..." → ✅ "A survey found that 73% of users..."
- ❌ "Significant improvements" → ✅ "Improvements of 23-47%"

**Test:** Can you back it up with a specific number, example, or name?

### 7. Too Many Transition Words

AI overuses transitions trying to seem sophisticated:

- ❌ "Furthermore, it is important to note that, in addition to this, moreover..."
- ✅ Strong structure needs fewer transitions

**Limit transitions to:** but, and, so, then, because

### 8. Overcomplicated Sentence Structure

**AI loves long, complex sentences trying to sound smart.**

- ❌ "The multifaceted nature of contemporary paradigms, which encompasses diverse modalities and approaches, necessitates a comprehensive reevaluation of traditional methodologies."
- ✅ "Things have changed. We need to rethink our approach."

---

## The Humanization Workflow

### Step 1: Understand Your Source

- What's the core insight or story?
- What's the emotional arc?
- What's the most important takeaway?

**Ask yourself:**
- [ ] What's being said that no one else is saying?
- [ ] What will readers think, feel, or do differently?
- [ ] What's the 30-second version?

### Step 2: Apply the SUCKS Framework

- [ ] Who is my ONE reader?
- [ ] Is this unique and useful to them?
- [ ] Can I make it clear, curious, and conversational?
- [ ] How do I keep it simple and structured?
- [ ] What will stick with them?

### Step 3: Write or Edit the Draft

If writing fresh:
1. Write freely without self-editing
2. Get the ideas down first
3. Polish in subsequent passes

If editing AI output:
1. Read the entire draft
2. Identify the core message
3. Rewrite rather than patch

### Step 4: Apply Sticky Sentences

Identify 2-3 most important statements and strengthen them:
- Add alliteration, symmetry, contrast, or rhythm
- Make them quotable
- Make them memorable

### Step 5: Eliminate AI Tells (Critical)

Go through systematically and remove:
- [ ] Correlative constructions ("X aren't just Y, they're Z")
- [ ] Overuse of "just" and "actually"
- [ ] Hedge words (might, could, perhaps, seems)
- [ ] Passive voice
- [ ] Corporate jargon
- [ ] Vague language (replace with specifics)
- [ ] Forbidden patterns ("Let that sink in," etc.)
- [ ] Too many transitions
- [ ] Overcomplicated sentences

**Test for each sentence:** Would this sentence appear in a ChatGPT output? If yes, rewrite it.

### Step 6: Read Aloud & Quality Check

Read your entire piece aloud:
- [ ] Does it sound like a real person talking?
- [ ] Does energy transfer (not feel formal)?
- [ ] Is every sentence earning its place?
- [ ] Do important ideas have sticky phrasing?
- [ ] Are there obvious AI tells remaining?

**If you answer "no" to any checkbox, fix it before publishing.**

---

## Quick Reference Checklists

### The "Would a Human Write This?" Test

For each paragraph, ask:
1. Does it sound like how I actually talk?
2. Is there unnecessary formality?
3. Are there words I'd never use in conversation?
4. Does it take the long way to make a point?
5. Does it hedge when I'm actually certain?

### The AI Tell Scan

Before publishing, scan for:
- [ ] "It's not X, it's Y" constructions
- [ ] "Just" appearing more than once
- [ ] "Actually" as sentence filler
- [ ] Passive voice ("was determined," "has been shown")
- [ ] Hedge words ("might," "could," "perhaps")
- [ ] Corporate speak ("leverage," "optimize," "ecosystem")
- [ ] Vague claims without specifics
- [ ] Overused transitions ("furthermore," "moreover")
- [ ] Complex sentences that could be simple
- [ ] Rhetorical flourishes ("The best part?")

### Elements of Effective Content

Include in your writing:
1. **Immediate attention** - First sentence hooks
2. **Specific numbers** - "1.5 million" not "many"
3. **Pattern interrupts** - Surprise the reader
4. **Direct problem-addressing** - Name the struggle
5. **Confidence** - Eliminate hedge words
6. **Concrete benefits** - "Double output in 30 days" not "boost productivity"
7. **Powerful questions** - Make readers think
8. **Warnings/stakes** - Create urgency

---

## Bundled Resources

### References
- `references/sticky-sentences.md` - Extended guide to memorable phrasing
- `references/forbidden-patterns.md` - Comprehensive list of AI tells
- `references/human-desires.md` - 8 core desires that drive engagement

---

## Related Skills

- **voice-analyzer** - Create a custom voice style from writing samples
- **voice-[style]** - Write in a specific author's voice
- **transcript-polisher** - Clean spoken content while preserving authenticity
- **hook-and-headline-writing** - Craft attention-grabbing headlines

---

*This skill is the foundation of the writing quality system. Use it standalone for basic humanization, or combine with voice styles for authentic, distinctive content.*

Related Skills

clear-writing

8
from cdeistopened/skill-stack

Write clear, human prose. Strips AI patterns, enforces hard constraints, optionally applies a voice style. Use when writing or editing any prose, when asked to "make it sound human," "remove AI tells," "write clearly," or when producing any content that must not read as AI-generated. Load style references for specific voices (Cowen, Koe, Levine-Berry, Lewis, Pirate Wires, OpenEd, Charlie) or social formats (Trung Phan).

youtube-scriptwriting

8
from cdeistopened/skill-stack

Transform raw ideas and brain dumps into polished YouTube scripts through a structured checkpoint workflow. Use when the user wants to write a YouTube script, improve video retention, craft hooks, structure educational or entertainment content, or turn source material (transcripts, notes, research) into a compelling video script. Guides through research, hook writing, story structure, body content, and editing phases.

writing-style

8
from cdeistopened/skill-stack

Core writing standards for all Skill Stack content. Invoke automatically for any writing task. Combines anti-AI patterns, voice principles, and prose style.

hook-and-headline-writing

8
from cdeistopened/skill-stack

Create attention-grabbing hooks and headlines using 15 proven formulas, sticky techniques, and the 4 U's test. This skill provides systematic frameworks for generating volume (10+ options) and selecting the best performers. Use for newsletter subject lines, social media hooks, blog headlines, or any content that needs to stop the scroll.

x-viral-template-miner

8
from cdeistopened/skill-stack

When the user wants to find proven-to-travel post templates in their niche and adapt them to their own product. Also use when the user mentions "what's going viral in my space", "what are competitors posting", "copy a viral post", "trending on X", "post ideas", "template mining", or "what to post this week". This is trend hunting, not plagiarism — the output is a template the user fills with their own assets.

x-linkedin-content-relay

8
from cdeistopened/skill-stack

When the user has X (Twitter) content that performed well and wants to relay it to LinkedIn 1-2 weeks later with reframing. Also use when the user mentions "repost to LinkedIn", "LinkedIn version of my tweet", "X to LinkedIn", "delayed repost", "LinkedIn for non-tech audience", or "LinkedIn relay". Also use when the user's ICP is non-tech and X is secondary — LinkedIn is the primary channel and this skill produces the content.

x-launch-video-structure

8
from cdeistopened/skill-stack

When the user is planning, scripting, or editing a product launch video for X (Twitter) and needs the structure. Also use when the user mentions "launch video", "demo video", "product launch on X", "60 second demo", "how to structure a launch", or "my launch video isn't working". Produces a beat-by-beat timing sheet, not copy.

x-account-warmup

8
from cdeistopened/skill-stack

When a user wants to grow an X (Twitter) account from zero before a product launch, or asks how to get first followers, warm up the algorithm, hit ~500-1,000 followers, or prepare an account to make a launch video land. Also use when the user mentions "new X account", "warm up my Twitter", "first 1000 followers", "building in public strategy", "X growth", or "engagement before launch".

skill-stack-thumbnails

8
from cdeistopened/skill-stack

Generate blog post thumbnails for Skill Stack using the brand aesthetic. Follows an iterative workflow - brainstorm concepts, get approval, generate with Gemini API.

youtube-ingest

8
from cdeistopened/skill-stack

Transcribe YouTube videos and playlists using Gemini Flash

web-scrape

8
from cdeistopened/skill-stack

Scrape web pages to clean markdown with optional AI summaries

voice-tyler-cowen

8
from cdeistopened/skill-stack

Write in Tyler Cowen's style - matter-of-fact, understated, treats enormous ideas as obvious observations. Read the passages. Absorb the flatness. Channel the HOW, not the content.