v2.0.0 to v2.1.0

30 added, 8 removed. Audit A to A.

---

name: human-first-writer

- description: "Use when the user asks to write authoritative, evidence-backed, human-first SEO content, articles, service pages, or local landing pages. Grounded in the 30-20-20-15-10-5 Editorial Weighting Model: Research Accuracy (30%), Search Intent (20%), Specificity (20%), Natural Flow (15%), SEO Entities (10%), and Subtle Cleanup (5%). Automatically outputs deliverables into markdown (.md) files with Title, Meta Description, Focus Keyword, and primary source citations. Trigger on: write human content, human-first writing, natural SEO article, write without sounding like AI, evidence-backed content."

+ description: "Use when the user asks to write authoritative, evidence-backed, human-first SEO content, articles, service pages, or local landing pages. Grounded in the 10-Stage Pipeline: Research → Verify → Search Intent → Local Context → Outline → Natural Draft → SEO → AI-pattern Cleanup → Fact Check → Final Human Edit. Automatically outputs deliverables into markdown (.md) files with Title, Meta Description, Focus Keyword, and verified data citations. Trigger on: write human content, human-first writing, natural SEO article, write without sounding like AI, evidence-backed content."

metadata:

- version: "2.0.0"

+ version: "2.1.0"

author: "Antigravity AI"

discipline: "content-engineering"

---

# Human-First Content Writer (The Evidence-First Editorial System)

A production-grade editorial system engineered to produce authoritative, high-converting, and genuinely natural content. Rather than relying on rigid mechanical AI evasion, this skill focuses on **empirical verification, verifiable sources, localized operational realities, and natural cadence**.

---

+ ## 🎯 The 10-Stage Production Pipeline

+ ```

+ [1. Deep Research] ➔ [2. Data Verification] ➔ [3. Intent Resolution]

+ │

+ ▼

+ [4. First-Hand/Local Context] ➔ [5. Structural Outline] ➔ [6. Natural Voice Draft]

+ │

+ ▼

+ [7. SEO & Entity Topology] ➔ [8. Subtle AI-Pattern Polish]

+ │

+ ▼

+ [9. Fact & Citation Check] ➔ [10. Final Human Editorial Edit]

+ ```

+ ### 🧠 Core Mindset Shift

+ - ❌ **Do NOT think:** "How do I avoid sounding like an AI?" (leads to over-engineered, unnatural text).

+ - ✅ **DO think:** "How would a seasoned, 10-year domain expert explain this with real facts and natural voice?"

+ ---

## ⚖️ The 6-Tier Editorial Priority Framework

```

┌────────────────────────────────────────────────────────┐

│ 1. Research Accuracy & Citation Integrity (30%) │

├────────────────────────────────────────────────────────┤

│ 2. Search Intent & Direct Value (20%) │

├────────────────────────────────────────────────────────┤

│ 3. Granular Specificity & Real Mechanisms (20%) │

├────────────────────────────────────────────────────────┤

│ 4. Natural Voice & Organic Flow (15%) │

├────────────────────────────────────────────────────────┤

│ 5. Technical & Entity-Driven SEO (10%) │

├────────────────────────────────────────────────────────┤

│ 6. Subtle AI-Slop Cleanup (5%) │

└────────────────────────────────────────────────────────┘

```

- ### 1. Research Accuracy & Citation Integrity (30%) — The Core Anchor

- - **No Floating Statistics:** Never state a percentage, timeline, or ranking without naming the institution, survey, or source (e.g. Bangladesh Bureau of Statistics, WHO Database, municipal gazettes).

- - **Nuanced Realism over Absolute Claims:** Replace sweeping absolutes ("100% zero waste", "zero traffic") with actual operational parameters (e.g. "implemented across core municipal wards", "significantly lower peak-hour gridlock along the primary arterial corridor").

- - **Mandatory Sources Section:** Conclude articles with named public references and methodology notes, NEVER an internal AI audit scorecard.

+ ### 1. Research Accuracy & Real Verification (30%)

+ - **No Citation by Name Only:** Mentioning "WHO" or "BBS" without specific survey years, report names, or tables is unacceptable. Ground metrics in verifiable baselines (e.g., *BBS HIES 2022 Divisional Housing Data*, *WHO 2016 Ambient Air Pollution Database Table 1*).

+ - **No Pseudo-Medical or Clinical Claims:** Never claim lifestyle changes "lower blood pressure" or "cure anxiety" without specific clinical studies. Use defensible environmental and lifestyle metrics (*cleaner air, shorter commutes, quieter residential areas*).

+ - **Geographic & Physical Grounding:** Explain the physical mechanism behind the numbers (e.g. commute times explained by metropolitan land area and arterial road corridors, not arbitrary guesses).

### 2. Search Intent Resolution (20%)

- Deliver the **Direct Answer Block** in the first 75-100 words. The reader must get their primary answer immediately without introductory throat-clearing.

### 3. Granular Specificity & Mechanisms (20%)

- Replace vague adjectives with exact materials, tools, official standards, or historical dates.

- For local content: Name specific street intersections, wards, regulatory bodies, and regional microclimates.

### 4. Natural Voice & Organic Flow (15%)

- Do not artificially ban natural grammar constructs. Writers use em-dashes and compound sentences when rhythm demands it.

- Focus on **Burstiness and Voice Variation**: Mix punchy analytical statements with detailed, multi-clause explanations.

### 5. Entity-Driven SEO (10%)

- Prepend the document with Title, Meta Description, Primary Focus Keyword, and LSI entities.

- Structure headings (`H2`, `H3`) to match topical search journeys.

### 6. Subtle AI-Slop Cleanup (5%)

- Avoid hollow AI buzzwords (`delve`, `tapestry`, `pivotal`, `paramount`, `elevate`, `seamless`, `in today's fast-paced world`).

- Cut empty introductory setup lines and generic "In conclusion" recaps.

---

## 📌 Deliverable Output Contract

Every content generation MUST:

1. Deliver the final copy into a dedicated `.md` file in the project workspace.

- 2. Include the metadata frontmatter at the top (Title, Meta, Focus Keyword).

- 3. Conclude with a verified **Sources & References** section.

+ 2. Include the metadata frontmatter at the top:

+ - `title`, `meta_description`, `focus_keyword`, `secondary_keywords`, `search_intent`, `word_count`, `file_path`.

+ 3. Conclude with a verified **Primary Sources, Municipal Data & References** section containing exact report titles and years.

4. **NEVER** append internal AI-validation tables or robotic checklists to the public article.