---
name: pulsedraft
description: Apply controlled entropy adjustments to AI-generated text to reduce statistical uniformity and make content feel more natural, human, and less robotic. Use when the user requests humanising AI text.
version: 1.0
---

# PulseDraft Skill

## Purpose

Transform AI-generated text into content that exhibits natural human variability while preserving:

- Factual accuracy  
- Core meaning  
- SEO intent  
- Structural clarity  
- Entity alignment

The goal is controlled irregularity, not randomness.

Do **NOT** introduce:

- False information  
- Fabricated examples  
- Artificial grammar mistakes  
- Slang exaggeration  
- Prompt leakage  
- Meta commentary

Output *only* the revised text.

---

## Activation Criteria

Activate this skill when:

- The user asks to humanise or enhance text tone
- The user wants text to feel less robotic
- The user requests reduced AI detectability
- The user wants natural rhythm improvements

Do **NOT** activate if:

- The task is purely grammar correction
- The task is translation
- The task is summarisation
- The task is SEO keyword insertion only

---

## Operating Model

Follow this 6-step process internally:

1. Read the full text  
2. Evaluate entropy weaknesses  
3. Identify dominant uniformity signals  
4. Apply selective revisions  
5. Re-assess coherence  
6. Output final text only

No explanation of edits.

---

## Entropy Optimization Pillars

### 1. Lexical Diversity

- Replace repetitive adjectives
- Introduce precise terminology
- Avoid stock SEO clichés
- Mix common and uncommon expressions

### 2. Syntactic Burstiness

- Vary sentence lengths
- Mix short and longer sentences
- Allow natural fragments
- Avoid rhythmic uniformity

### 3. Semantic Flow Variation

- Introduce subtle reflective pivots
- Permit mild contextual digressions
- Return cleanly to the main thread

### 4. Transition Reduction

- Minimise overuse of mechanical transitions like:
  - However
  - Moreover
  - In addition
  - Therefore
  - Ultimately
- Prefer implicit flow where natural

### 5. Emotional Micro-Variance

- Add mild perspective shifts
- Avoid flat informational monotone

### 6. Formatting Asymmetry

If lists are present:

- Vary structure lengths
- Avoid perfect symmetry
- Mix prose and bullets when helpful

### 7. Passive Voice Control

Prefer active voice, use passive where contextually appropriate.

### 8. Cliché Elimination

Remove phrases like:

- In today’s world
- Game changer
- Cutting edge

Replace with precise language.

### 9. Nuance Calibration

- Avoid absolute claims
- Use conditional phrasing where appropriate
- Preserve credibility

### 10. Perplexity Volatility

Increase statistical variability without reducing clarity.

---

## Output Rules

- Output only the revised text  
- Do not include commentary  
- Do not mention AI  
- Do not describe edits  
- Preserve headings unless structural improvements are required  
- Maintain original meaning

---

## Over-Optimization Warning

Do **NOT**:

- Inject randomness for randomness’ sake  
- Degrade readability  
- Change core message  
- Inflate tone unnecessarily  
- Insert false specificity

The goal is natural human rhythm.

---

## Good vs Bad Examples

Below are illustrative examples showing how PulseDraft should transform text.

### Example 1: Repetition

**Bad:**  
The product was very effective because it was very efficient and very reliable. It was very well designed and very easy to use.

**Good:**  
The product proved effective, with reliable performance and a design that made it easy to use.

**Why:** The bad version repeats “very” and similar adjectives; the good version uses precise language and varied structure.

---

### Example 2: Robotic Transitions

**Bad:**  
Moreover, the system improves performance. In addition, it reduces cost. Therefore, users benefit overall.

**Good:**  
The system improves performance while cutting costs, giving users a clear advantage.

**Why:** The bad version uses mechanical transitions; the good version flows naturally without redundant connectors.

---

### Example 3: Burstiness and Rhythm

**Bad:**  
The launch was successful. It exceeded expectations. It demonstrated strong adoption across markets. It met every target.

**Good:**  
The launch exceeded expectations, showing strong adoption across markets and meeting every target.

**Why:** Short, repetitive sentences create a monotone rhythm; the good version mixes structures for better flow.

---

### Example 4: Clichés and Filler

**Bad:**  
In today’s world, there are many challenges that make it hard for companies to innovate and succeed in every market.

**Good:**  
Companies face structural and competitive hurdles that complicate innovation and market success.

**Why:** The bad version uses “in today’s world” (a cliché) and general phrasing; the good version is specific and direct.

---

### Example 5: Passive vs Active Voice

**Bad:**  
The report was read by the entire team before the deadline was met.

**Good:**  
The team read the report before the deadline.

**Why:** Active voice is clearer, stronger, and feels more human.  

---

## Success Criteria

The final output should:

- Exhibit uneven, natural rhythm  
- Contain varied sentence lengths  
- Avoid mechanical transitions  
- Maintain coherence  
- Preserve factual integrity  
- Reduce detectable uniformity
- Preserve factual integrity.
- Reduce detectable AI uniformity.

End of Skill.