Thesis Reviewer — Systematic Thesis Review for All Disciplines

中文文档

An AI coding agent skill for systematically reviewing master's and doctoral theses across all academic disciplines. Designed for both supervisors and students — supervisors use it to generate structured review feedback, students use it for pre-submission self-checks to catch issues before blind review.

Who is it for?

What it does

• Converts .docx thesis to Markdown via markitdown MCP, then reviews chapter by chapter
• Supports all disciplines with 7 discipline-specific review modules + universal framework
• Supports both master's and PhD theses — auto-detects degree type, adapts review standards
• Academic vs professional degree distinction per 2025 Degree Law
• 5 review dimensions: academic quality, writing quality, formatting (GB/T 7713.1), data & results, academic integrity
• Two-phase workflow: Phase 1 auto deep analysis → Phase 2 interactive refinement
• 149 universal checklist items + 22-34 discipline-specific supplements (170+ per review)
• PhD-specific evaluation: originality assessment, independent research capability, research system coherence, publication record
• Blind review risk assessment: predicts potential issues across 5 review dimensions
• Severity markers: 🔴 Serious / 🟡 Needs improvement / 🟢 Good — no ambiguous numeric scores
• Cross-chapter consistency checks: questions↔results, methods↔data, citations↔references
• Prioritized revision roadmap — students know exactly what to fix first
• National standards aligned: GB/T 7713.1-2006, GB/T 7714-2015, GB 3100-3102-93

Multi-Platform Support

Works with all major AI coding agents that support the Agent Skills format:

Comparison

vs No Skill (native agent)

How it works

Input .docx
    │
    ▼
[Preprocessing] markitdown → Markdown → identify chapter structure
    │
    ▼
[Phase 1: Auto Deep Analysis]
    ├─ Step 1: Overall scan (structure, research question, global impression)
    ├─ Step 2: Chapter-by-chapter + global analysis (170+ checklist items)
    ├─ Step 2b: Cross-chapter consistency checks
    └─ Step 3: Generate draft review report
    │
    ▼
[Phase 2: Interactive Refinement]
    ├─ Discuss specific chapters in depth
    ├─ Ask follow-up questions about specific issues
    ├─ Adjust severity markers, add/remove opinions
    └─ User says "完成精修" to finalize
    │
    ▼
[Final] Merge all changes → save final review report as Markdown

Review Dimensions

1. Academic Quality

Abstract, introduction/literature review, materials & methods, results, discussion, conclusion, innovation — with domain-specific checks for experimental design, sample size, controls, and biological replicates.

2. Writing Quality

Logical coherence between chapters, argumentative rigor (claim-evidence-reasoning), academic language standards, abstract quality (Chinese + English).

3. Formatting

Chapter completeness, figure/table standards (numbering, titles, resolution, three-line tables), reference format consistency, abbreviations, biological nomenclature (gene names in italics).

4. Data & Results

Appropriate chart types, error bars/confidence intervals, statistical test selection (parametric vs non-parametric), p-value notation, multiple comparison correction, figure quality (axis labels, legends, color-blind friendly), reproducibility.

5. Academic Integrity

Plagiarism detection, image manipulation, data fabrication, citation ethics, originality declaration completeness.

Discipline-Specific Modules

PhD-Specific Dimensions (additional)

• Originality: original contribution vs incremental improvement, publishability at high-impact journals/conferences
• Independent research capability: ability to formulate questions, design research, diagnose problems
• Research system coherence: logical connections between multi-chapter studies, progressive depth
• Literature review depth: field-level understanding, coverage of classic and cutting-edge work, critical evaluation of debates
• Publication record: papers published/accepted during candidacy, relationship to thesis content

Prerequisites

• markitdown MCP — required for .docx to Markdown conversion

Skill Installation

Claude Code

# Global install (available in all projects)
git clone https://github.com/Agents365-ai/thesis-reviewer.git ~/.claude/skills/thesis-reviewer

# Project-level install
git clone https://github.com/Agents365-ai/thesis-reviewer.git .claude/skills/thesis-reviewer

OpenClaw / ClawHub

# Via ClawHub
clawhub install thesis-reviewer

# Manual install
git clone https://github.com/Agents365-ai/thesis-reviewer.git ~/.openclaw/skills/thesis-reviewer

# Project-level install
git clone https://github.com/Agents365-ai/thesis-reviewer.git skills/thesis-reviewer

Hermes Agent

# Install under research category
git clone https://github.com/Agents365-ai/thesis-reviewer.git ~/.hermes/skills/research/thesis-reviewer

Or add an external directory in ~/.hermes/config.yaml:

skills:
  external_dirs:
    - ~/myskills/thesis-reviewer

OpenAI Codex

# User-level install
git clone https://github.com/Agents365-ai/thesis-reviewer.git ~/.agents/skills/thesis-reviewer

# Project-level install
git clone https://github.com/Agents365-ai/thesis-reviewer.git .agents/skills/thesis-reviewer

OpenCode

# Global install
git clone https://github.com/Agents365-ai/thesis-reviewer.git ~/.opencode/skills/thesis-reviewer

Pi-Mono

# Global install
git clone https://github.com/Agents365-ai/thesis-reviewer.git ~/.pimo/skills/thesis-reviewer

SkillsMP

skills install thesis-reviewer

Installation paths summary

Usage

Just provide your thesis file:

Review this master's thesis: /path/to/thesis.docx

Or in Chinese:

帮我评审这篇硕士论文：/path/to/thesis.docx

The skill will automatically convert, analyze, and generate a structured review report.

Trigger Keywords

• Chinese: 论文评审, 学位论文, 审阅论文, 论文修改意见, 硕士论文, 博士论文, 毕业论文
• English: thesis review, dissertation review, thesis feedback, PhD thesis, doctoral thesis

Output Files

Files

• SKILL.md — main skill instructions, loaded by all platforms
• checklist.md — universal review checklist across 5 dimensions (loaded by SKILL.md during review)
• output-template.md — review report output template (loaded by SKILL.md for report generation)
• disciplines/ — discipline-specific review modules:
  • life-sciences.md — biology, biomedical, ecology, agriculture
  • medicine.md — clinical medicine, public health, pharmacy
  • cs-ai.md — computer science, AI, machine learning
  • engineering.md — mechanical, electrical, chemical, civil engineering
  • chemistry.md — chemistry, materials science
  • physics.md — theoretical, experimental, computational physics
  • social-sciences.md — economics, management, law, education, psychology
• agents/openai.yaml — OpenAI Codex-specific configuration
• README.md — this file (English, displayed on GitHub homepage)
• README_CN.md — Chinese documentation

Note: SKILL.md, checklist.md, output-template.md, and disciplines/ are needed for the skill to work. agents/openai.yaml is only needed for Codex. The README files are documentation only.

Known Limitations

• Requires markitdown MCP: The skill depends on the markitdown MCP tool for .docx conversion. Without it, you must manually convert the thesis to Markdown first.
• Long documents: Very long theses (>80,000 words) may require multiple reading passes due to context window limits. The chapter-by-chapter approach mitigates this.
• Formatting detection: Some formatting issues (image resolution, page headers) cannot be fully detected from the Markdown conversion and may need manual checking.
• Language: Review output is in Simplified Chinese, designed for Chinese university thesis review workflows.

License

MIT

Support

If this skill helps you, consider supporting the author:

WeChat Pay

Alipay

Buy Me a Coffee

Author

Agents365-ai

• Bilibili: https://space.bilibili.com/441831884
• GitHub: https://github.com/Agents365-ai