---
name: academic-writing
description: "General-purpose academic writing framework for all disciplines — literature review, thesis development, structure templates, citation management, quality gates. Works alongside openalex and arxiv for any field."
version: 1.0.0
author: Hermes Agent
license: MIT
platforms: [linux, macos, windows]
metadata:
  hermes:
    tags: [Academic, Writing, Literature Review, Citation, Research, Thesis, Structure, Quality]
    category: research
    related_skills: [openalex, arxiv, research-paper-writing]
    requires_toolsets: [terminal, web]
---

# Academic Writing — General Framework

Universal academic writing workflow: from research question → literature review → structure → drafting → quality assurance. Discipline-agnostic — applies to humanities, social sciences, engineering, AI, and cross-disciplinary work.

## When To Use

- Writing a literature review, thesis chapter, research article, or dissertation
- Structuring an interdisciplinary paper across multiple fields
- Evaluating or improving existing academic writing
- Preparing a paper for a specific journal or conference
- Translating academic work between Chinese and English

## Core Principle

**AI is your copilot, not the pilot.** AI handles: hunting references, formatting citations, checking consistency, detecting logical gaps. **Human owns:** defining research questions, choosing methods, interpreting evidence, crafting arguments.

## Anti-Hallucination Rules (Always Enforce)

These apply to ALL drafts, regardless of discipline:

| Rule | Description |
|------|-------------|
| **R1** | Never invent or approximate citations. If you don't know the exact author, title, year, and source — don't cite it. |
| **R2** | Never cite a paper in a year that hasn't happened yet. |
| **R3** | When citing a claim, ensure the cited source actually supports that specific claim, not just the general topic. |
| **R4** | For empirical claims (statistics, experimental results), verify the number appears in the source. |
| **R5** | Never use "many scholars argue", "research shows", "it is well known" without a specific citation. |

## Research Question → Thesis

### Socratic Questioning (Before Writing Anything)

Work through these in order:

1. **Interest**: What specific phenomenon/tension/existing claim are you investigating?
2. **Gap**: What does existing literature NOT explain, overlook, or get wrong?
3. **Contribution**: What is your specific new argument/evidence/framework?
4. **Scope**: What is explicitly NOT covered by your paper?
5. **Stakes**: Why should anyone in the field care about this?

Output: A **thesis statement** (1-2 sentences) that can be defended in the paper's length.

## Literature Review Methodology

### Finding Sources (Use Skills)

| Field | Primary Skill | Backup |
|-------|--------------|--------|
| CS/AI/ML | `arxiv` | Semantic Scholar |
| Engineering/Energy | `openalex` (filter D4) | Google Scholar |
| Philosophy | `openalex` (filter D2) | PhilPapers |
| Art/Visual Studies | `openalex` (filter D3) | JSTOR |
| Social Sciences | `openalex` (filter D1) | Google Scholar |
| Mixed/Interdisciplinary | `openalex` + `arxiv` combined | — |

### Source Evaluation Checklist

For each source, assess:
- [ ] What is the author's **central claim** in one sentence?
- [ ] What **evidence** supports it?
- [ ] What **limitation** does the author acknowledge?
- [ ] How does this **connect to or challenge** your thesis?
- [ ] What **upstream** work does this author build on? (check their references)

### Source Integration (5-Layer Method)

From `humanities-thesis-skill` approach:

| Layer | Question |
|-------|----------|
| 1. Meta | Who wrote this, when, in what context? |
| 2. Argument | What is the MAIN argument (not summary)? |
| 3. Evidence | What specific evidence is used? |
| 4. Positioning | How does this relate to your thesis? |
| 5. Cross-check | Does this agree or conflict with other sources? |

## Paper Structure Templates

### Standard Research Article (Any Discipline)

```
1. Title — specific, not generic
2. Abstract — problem, method, finding, implication (4 sentences)
3. Introduction
   3.1 Research context / problem statement
   3.2 Gap identification (what existing work misses)
   3.3 Thesis / contribution statement
   3.4 Scope and boundaries
   3.5 Paper roadmap
4. Literature Review / Related Work
   4.1 Thematic organization (NOT source-by-source)
   4.2 Synthesis: show where your work fits
5. Methodology / Theoretical Framework
   5.1 For empirical: data sources, methods, limitations
   5.2 For humanities: theoretical lens, source base, interpretive method
6. Analysis / Argument Development
7. Discussion
   7.1 What the findings mean
   7.2 How they confirm/challenge existing work
   7.3 Limitations
   7.4 Future directions
8. Conclusion
9. References
```

### Literature Review Article

```
1. Introduction — scope, organization rationale, thesis
2. Thematic sections — organize by theme/debate, NOT by source
3. Synthesis — identify patterns, disagreements, gaps
4. Conclusion — what we know, what remains unknown, why it matters
```

### Humanities / Interpretive Essay

```
1. Introduction — specific interpretive claim, not topic
2. Close reading / Primary source analysis
3. Theoretical framing — apply relevant theory to evidence
4. Contextualization — historical, cultural, intellectual
5. Discussion — broader implications of interpretation
6. Conclusion
```

### Engineering / Technical Paper

```
1. Abstract — problem, approach, key results, significance
2. Introduction — problem, motivation, related work, contribution
3. Background / Preliminaries
4. Proposed Method / System Design
5. Evaluation (experiments, analysis, proofs)
6. Discussion / Limitations
7. Related Work
8. Conclusion
```

## Citation Management

### Citation Style Quick Reference

| Style | Use When |
|-------|----------|
| APA 7th | Social sciences, education, psychology |
| IEEE | Engineering, computer science |
| Chicago | History, humanities (notes-bibliography) |
| MLA | Literature, arts, cultural studies |
| Harvard |通用人文社科 |

### Generating Citations

```bash
# BibTeX from DOI (CrossRef)
curl -s "https://api.crossref.org/works/10.1038/nature12373" | python3 -c "
import sys,json
d=json.load(sys.stdin)['message']
print(f\"@article{{{d['author'][0]['family']}{d['published-print']['date-parts'][0][0]},
  title={{{d['title'][0]}}},
  author={{{' and '.join(a['family'] for a in d['author'])}}},
  journal={{{d['container-title'][0]}}},
  year={{{d['published-print']['date-parts'][0][0]}}},
  volume={{{d.get('volume','')}}},
  pages={{{d.get('page','')}}}
}}\")"
```

## Quality Gates

### Before Drafting
- [ ] Research question is specific and answerable
- [ ] At least 15-20 core sources identified (more for dissertations)
- [ ] Gap/contribution is clearly stated
- [ ] Paper structure fits the argument type

### During Drafting
- [ ] Every claim has a specific citation (no vague "scholars argue")
- [ ] Each paragraph has ONE main idea, stated in the first sentence
- [ ] Transitions connect paragraphs logically
- [ ] Evidence paragraphs: claim → evidence → analysis → link back to thesis
- [ ] Theory sections: apply theory to evidence, don't just describe theory

### After Drafting
- [ ] Abstract: reads as a complete micro-paper (all 4 elements present)
- [ ] Introduction: someone could guess the paper's structure from it
- [ ] Conclusion: no new claims introduced
- [ ] References: all in-text citations appear in reference list
- [ ] Word count: within target journal/conference limits

## Writing Bilingual (Chinese ↔ English Academic)

### Translation Principles
- Translate meaning, not word-for-word
- Chinese academic writing tends to be more indirect; English tends to be direct
- Preserve technical terminology — use established translations from the field
- For philosophy terms (e.g., 延异/différance), keep French original in parentheses first time

### English Abstract Structure
```
[Context 1 sentence] + [Gap/Problem 1 sentence] + [Method 1 sentence] + [Finding 1 sentence] + [Implication 1 sentence] = 5 sentences
```

### Chinese Abstract Structure
```
[研究背景/问题] + [研究目的/方法] + [主要发现/论点] + [研究意义] = 4-5 sentences
```

## Common Failure Modes

| Problem | Diagnosis | Fix |
|---------|-----------|-----|
| "So what?" feedback | Gap/contribution not in intro | Rewrite intro to foreground contribution |
| "Too descriptive" | Literature review is summary, not synthesis | Organize by themes/debates, not by paper |
| "Unfocused" | Too many themes, underdeveloped | Cut to 2-3 core themes, deepen each |
| "Sources don't support claims" | Claims too strong for sources | Qualify claims or find stronger sources |
| "Jargon overload" | Over-reliance on field-specific terms | Define all acronyms, explain specialist terms |
| "Chinese citations only" | One-sourced argument | Cross-check with international literature |

## Tools Reference

| Tool | Purpose |
|------|---------|
| `openalex` skill | Cross-disciplinary paper search |
| `arxiv` skill | CS/AI/ML papers |
| `ocr-and-documents` skill | PDF text extraction |
| `search_openalex.py` script | Batch search with filters |