Content Research WriterSAFE

Intelligent Skill routing and workflow orchestration for AI agents — +21.12 pp reward, −29.6% tokens on SkillsBench with DeepSeekV4Flash-VE.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: content-research-writer
description: Assists in writing high-quality content by conducting research, adding citations, improving hooks, iterating on outlines, and providing real-time feedback on each section. Transforms your writing process from solo effort to collaborative partnership.
---

# Content Research Writer

This skill acts as your writing partner, helping you research, outline, draft, and refine content while maintaining your unique voice and style.

## When to Use This Skill

- Writing blog posts, articles, or newsletters
- Creating educational content or tutorials
- Drafting thought leadership pieces
- Researching and writing case studies
- Producing technical documentation with sources
- Writing with proper citations and references
- Improving hooks and introductions
- Getting section-by-section feedback while writing

## What This Skill Does

1. **Collaborative Outlining**: Helps you structure ideas into coherent outlines
2. **Research Assistance**: Finds relevant information and adds citations
3. **Hook Improvement**: Strengthens your opening to capture attention
4. **Section Feedback**: Reviews each section as you write
5. **Voice Preservation**: Maintains your writing style and tone
6. **Citation Management**: Adds and formats references properly
7. **Iterative Refinement**: Helps you improve through multiple drafts

## How to Use

### Setup Your Writing Environment

Create a dedicated folder for your article:
```
mkdir ~/writing/my-article-title
cd ~/writing/my-article-title
```

Create your draft file:
```
touch article-draft.md
```

Open Claude Code from this directory and start writing.

### Basic Workflow

1. **Start with an outline**:
```
Help me create an outline for an article about [topic]
```

2. **Research and add citations**:
```
Research [specific topic] and add citations to my outline
```

3. **Improve the hook**:
```
Here's my introduction. Help me make the hook more compelling.
```

4. **Get section feedback**:
```
I just finished the "Why This Matters" section. Review it and give feedback.
```

5. **Refine and polish**:
```
Review the full draft for flow, clarity, and consistency.
```

## Instructions

When a user requests writing assistance:

1. **Understand the Writing Project**
   
   Ask clarifying questions:
   - What's the topic and main argument?
   - Who's the target audience?
   - What's the desired length/format?
   - What's your goal? (educate, persuade, entertain, explain)
   - Any existing research or sources to include?
   - What's your writing style? (formal, conversational, technical)

2. **Collaborative Outlining**
   
   Help structure the content:
   
   ```markdown
   # Article Outline: [Title]
   
   ## Hook
   - [Opening line/story/statistic]
   - [Why reader should care]
   
   ## Introduction
   - Context and background
   - Problem statement
   - What this article covers
   
   ## Main Sections
   
   ### Section 1: [Title]
   - Key point A
   - Key point B
   - Example/evidence
   - [Research needed: specific topic]
   
   ### Section 2: [Title]
   - Key point C
   - Key point D
   - Data/citation needed
   
   ### Section 3: [Title]
   - Key point E
   - Counter-arguments
   - Resolution
   
   ## Conclusion
   - Summary of main points
   - Call to action
   - Final thought
   
   ## Research To-Do
   - [ ] Find data on [topic]
   - [ ] Get examples of [concept]
   - [ ] Source citation for [claim]
   ```
   
   **Iterate on outline**:
   - Adjust based on feedback
   - Ensure logical flow
   - Identify research gaps
   - Mark sections for deep dives

3. **Conduct Research**
   
   When user requests research on a topic:
   
   - Search for relevant information
   - Find credible sources
   - Extract key facts, quotes, and data
   - Add citations in requested format
   
   Example output:
   ```markdown
   ## Research: AI Impact on Productivity
   
   Key Findings:
   
   1. **Productivity Gains**: Studies show 40% time savings for 
      content creation tasks [1]
   
   2. **Adoption Rates**: 67% of knowledge workers use AI tools 
      weekly [2]
   
   3. **Expert Quote**: "AI augments rather than replaces human 
      creativity" - Dr. Jane Smith, MIT [3]
   
   Citations:
   [1] McKinsey Global Institute. (2024). "The Economic Potential 
       of Generative AI"
   [2] Stack Overflow Developer Survey (2024)
   [3] Smith, J. (2024). MIT Technology Review interview
   
   Added to outline under Section 2.
   ```

4. **Improve Hooks**
   
   When user shares an introduction, analyze and strengthen:
   
   **Current Hook Analysis**:
   - What works: [positive elements]
   - What could be stronger: [areas for improvement]
   - Emotional impact: [current vs. potential]
   
   **Suggested Alternatives**:
   
   Option 1: [Bold statement]
   > [Example]
   *Why it works: [explanation]*
   
   Option 2: [Personal story]
   > [Example]
   *Why it works: [explanation]*
   
   Option 3: [Surprising data]
   > [Example]
   *Why it works: [explanation]*
   
   **Questions to hook**:
   - Does it create curiosity?
   - Does it promise value?
   - Is it specific enough?
   - Does it match the audience?

5. **Provide Section-by-Section Feedback**
   
   As user writes each section, review for:
   
   ```markdown
   # Feedback: [Section Name]
   
   ## What Works Well ✓
   - [Strength 1]
   - [Strength 2]
   - [Strength 3]
   
   ## Suggestions for Improvement
   
   ### Clarity
   - [Specific issue] → [Suggested fix]
   - [Complex sentence] → [Simpler alternative]
   
   ### Flow
   - [Transition issue] → [Better connection]
   - [Paragraph order] → [Suggested reordering]
   
   ### Evidence
   - [Claim needing support] → [Add citation or example]
   - [Generic statement] → [Make more specific]
   
   ### Style
   - [Tone inconsistency] → [Match your voice better]
   - [Word choice] → [Stronger alternative]
   
   ## Specific Line Edits
   
   Original:
   > [Exact quote from draft]
   
   Suggested:
   > [Improved version]
   

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.