| name | paper-review |
| description | Review academic papers in LaTeX projects. Analyzes TeX source files, figures, and bibliography to provide comprehensive peer review feedback. Use when the user asks to review a paper, evaluate a submission, or wants feedback on their LaTeX manuscript. |
| version | 1.0.0 |
| author | Academic Paper Tools |
| license | MIT |
| tags | ["Academic","Paper Review","LaTeX","Peer Review","NeurIPS","ICML","ICLR","CVPR"] |
Paper Review - AI-Powered Academic Paper Evaluation
This skill provides comprehensive peer review for LaTeX academic papers by directly analyzing source files.
Activation Triggers
Use this skill when the user:
- Asks to "review this paper" or "review my paper"
- Wants feedback on a LaTeX manuscript
- Asks about paper quality, novelty, or acceptance chances
- Mentions ICLR, NeurIPS, ICML, CVPR, or other venue submissions
Review Process
Step 1: Discover Project Structure
First, identify the LaTeX project structure:
Typical LaTeX project:
├── main.tex # Main document (look for \documentclass)
├── sections/ # Or mysec/, chapters/, etc.
│ ├── 1_intro.tex
│ ├── 2_method.tex
│ └── ...
├── figures/ # Or figs/, images/, asset/
│ ├── figure1.pdf
│ └── ...
├── main.bib # Or references.bib, ref.bib
└── *.sty # Style files (can ignore)
Actions:
- Use
LS to explore the project root
- Find the main
.tex file (contains \documentclass)
- Locate figure directories and
.bib files
Step 2: Read Core Content
Read files in this priority order:
- Main tex file - Get document structure, abstract, title
- Introduction - Understand motivation and contributions
- Method/Approach - Core technical content
- Experiments - Results and evaluation
- Figures - Visual quality and clarity (use
Read on image files)
- Bibliography - Reference quality and completeness
Tips:
- For
\input{sections/intro}, read sections/intro.tex
- For
\includegraphics{figures/fig1}, read the image file
- Pay attention to
\cite{} commands for reference analysis
Step 3: Evaluate Paper Quality
Assess the following dimensions:
3.1 Technical Contribution
- Novelty: Groundbreaking / Significant / Incremental
- Soundness: Are the methods technically correct?
- Completeness: Is the approach fully described?
3.2 Experimental Evaluation
- Baselines: Are comparisons fair and comprehensive?
- Datasets: Appropriate and sufficient?
- Metrics: Standard and meaningful?
- Ablations: Do they validate design choices?
3.3 Presentation Quality
- Clarity: Is the writing clear and well-organized?
- Figures: High quality, informative, readable?
- Math: Notation consistent? Equations well-explained?
3.4 Related Work
- Coverage: Are key papers cited?
- Positioning: How does this work differ from prior art?
Step 4: Generate Review Report
Provide a structured review in this format:
## Paper Review: [Title]
### Summary
[2-3 sentence summary of the paper's main contribution]
### Strengths
1. [Major strength with specific reference to content]
2. [Another strength]
3. ...
### Weaknesses
1. [Major weakness with specific location, e.g., "Section 3.2, line discussing X"]
2. [Another weakness]
3. ...
### Questions for Authors
1. [Clarifying question]
2. ...
### Detailed Comments
#### Introduction (sections/1_intro.tex)
- [Specific feedback with line references if helpful]
#### Method (sections/3_method.tex)
- [Specific feedback]
#### Experiments (sections/4_experiments.tex)
- [Specific feedback]
#### Figures
- Figure 1: [Assessment]
- Figure 2: [Assessment]
### Overall Assessment
| Aspect | Rating | Comment |
|--------|--------|---------|
| Novelty | ⭐⭐⭐☆☆ | [Brief comment] |
| Soundness | ⭐⭐⭐⭐☆ | [Brief comment] |
| Presentation | ⭐⭐⭐☆☆ | [Brief comment] |
| Significance | ⭐⭐⭐☆☆ | [Brief comment] |
**Recommendation**: Accept / Weak Accept / Borderline / Weak Reject / Reject
**Confidence**: High / Medium / Low
Special Modes
Quick Review Mode
If user asks for a "quick review" or "brief feedback":
- Read only main.tex, intro, and method sections
- Provide 3 strengths, 3 weaknesses, overall recommendation
Focus Review Mode
If user asks about specific aspects (e.g., "review my experiments"):
- Focus on the requested sections
- Provide deeper analysis of that specific area
Rebuttal Helper Mode
If user provides reviewer comments:
- Analyze each reviewer concern
- Suggest specific responses with references to paper content
- Identify which concerns need new experiments vs. clarification
Important Notes
- Be Constructive: Frame criticism as suggestions for improvement
- Be Specific: Reference exact locations (file, section, paragraph)
- Be Fair: Acknowledge both strengths and weaknesses
- Be Actionable: Provide concrete suggestions, not vague complaints
- Consider Venue: Adjust standards based on target venue (top-tier vs. workshop)
Example Interaction
User: Review my paper in the release/ folder
Agent Actions:
LS release/ → Find main.tex, mysec/, figures/, main.bib
Read release/main.tex → Get structure, abstract
Read release/mysec/1_intro.tex → Analyze introduction
Read release/mysec/3_method.tex → Analyze methodology
Read release/mysec/4_experiments.tex → Analyze experiments
Read release/figures/figure1.pdf → Check figure quality
- Synthesize findings into structured review report