content-humanizer

Files (4)

• SKILL.md
• ai_pattern_detector.py · scripts/
• content_scorer.py · scripts/
• readability_scorer.py · scripts/

SKILL.md (first lines)

# Content Humanizer

Transform machine-sounding content into writing that reads like it came from a real person with real opinions and real experience.

---

## Table of Contents

- [Keywords](#keywords)
- [Quick Start](#quick-start)
- [Core Workflows](#core-workflows)
- [AI Pattern Detection Catalog](#ai-pattern-detection-catalog)
- [Humanization Techniques](#humanization-techniques)
- [Voice Injection Framework](#voice-injection-framework)
- [Rhythm and Cadence Repair](#rhythm-and-cadence-repair)
- [Specificity Replacement Guide](#specificity-replacement-guide)
- [Before and After Examples](#before-and-after-examples)
- [Best Practices](#best-practices)
- [Integration Points](#integration-points)

---

## Keywords

content humanizer, AI content, humanize writing, AI detection, natural writing, authentic content, AI cliches, robotic writing, brand voice, personality injection, writing rhythm, AI patterns, content authenticity, human voice, AI tells, content polishing, voice consistency, writing style, content quality

---

## Clarify First

Before humanizing, settle the inputs below. When one is missing or vague and would change the result, ask about it; when it wouldn't, use a sensible default and say which one you used.

- **Mode:** audit-only (annotated report) or full humanization rewrite (sets the deliverable and how invasive edits are).
- **Content type:** docs, blog post, marketing copy, or email (determines how much personality to inject vs. preserve clarity).
- **Brand voice:** guidelines or one example of writing they love (without it, voice injection is guesswork).

Ask about the two or three inputs that matter most for this request, not the whole list. If the user wants a result straight away, produce it and state your assumptions at the top.

## Quick Start

### Detect AI Patterns in Content

1. Scan for overused filler words (delve, landscape, crucial, leverage, robust)
2. Check for hedging chains ("It's important to note that...")
3. Count em-dash frequency (more than 2 per 500 words = AI fingerprint)
4. Evaluate paragraph structure uniformity (identical patterns = AI)
5. Flag all unattributed vague claims ("Many companies," "Studies show")
6. Score severity: High (10+ tells per 500 words = full rewrite needed)

### Humanize a Draft

1. Replace all filler words with plain-language alternatives
2. Vary sentence length deliberately (short, long, medium, fragment)
3. Replace every vague claim with a specific data point or honest qualification
4. Break uniform paragraph structure with fragments, questions, and asides
5. Add friction and imperfection (qualifications, direction changes, opinions)
6. Inject brand voice if voice guidelines exist

---

## Core Workflows

### Workflow 1: AI Pattern Audit (Diagnostic Only)

Scan content without editing. Produce an annotated report.

**Step 1: Run Detection Scan**

Flag every instance in these categories with severity ratings:
- Critical (kills credibility): Overused filler words, hedging chains, identical paragraph structure, lack of specificity
- Medium (softens impact): Em-dash overuse, false certainty, generic conclusions
- Minor (polish only): Slightly repetitive transitions, mild formatting uniformity

**Step 2: Count and Score**

| Metric | Threshold |
|--------|-----------|
| AI tells per 500 words | < 3 = minor edits needed, 3-7 = significant editing, 8+ = full rewrite |
| Unique paragraph structures | < 3 patterns in 1,000+ words = AI fingerprint |
| Vague claims without attribution | Any = flag each one |
| Sentences starting with "It is" | > 3 per 1,000 words = flag |

**Step 3: Deliver Audit Report**

```markdown
## AI Pattern Audit
Content: [Title or description]
Word count: [X]
AI Tell Count: [X] (Critical: [X], Medium: [X], Minor: [X])
Recommendation: [Minor edits / Significant editing / Full rewrite]

### Critical Issues
[Each issue with line reference, pattern category, and specific fix]

### Medium Issues
[Same format]

### Minor Issues
[Same format]
```

### Workflow 2: Full Humanization Pass

Transform the content from AI-sounding to authentically human.

**Step 1: Remove AI Filler Words**

Never just delete — always replace with something better or restructure the sentence:

| AI Phrase | Replacement Options |
|-----------|-------------------|
| "delve into" | "look at," "dig into," "break down," or restructure without the phrase |
| "the [X] landscape" | "how [X] works today," "the current state of [X]" |
| "leverage" | "use," "apply," "put to work" |
| "crucial" / "vital" / "pivotal" | State the thing and let it be self-evidently important |
| "furthermore" / "moreover" | Start the next sentence directly, or use "and" or "also" |
| "robust" / "comprehensive" | Replace with specific description of what it actually covers |
| "facilitate" / "foster" | "help," "make easier," "allow," "create" |
| "navigate this challenge" | "handle this," "deal with this," "get through this" |
| "in order to" | "to" |

From borghei/Claude-Skills, MIT.

More from borghei/Claude-Skills