Claude-skill-registry add-paper
Add a research paper to the maxpool research-papers collection. Use when the user provides an ArXiv URL, PDF link, or asks to add/summarize a research paper for the website. Handles paper fetching, insight extraction, HTML generation, and index updates.
git clone https://github.com/majiayu000/claude-skill-registry
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/add-paper" ~/.claude/skills/majiayu000-claude-skill-registry-add-paper && rm -rf "$T"
skills/data/add-paper/SKILL.mdAdd Research Paper Skill
This skill automates adding research papers to the
/research-papers/ directory following the established template patterns.
When to Use This Skill
Trigger this skill when the user:
- Provides an ArXiv URL (html, abs, or pdf format)
- Asks to "add a paper" or "create a research summary"
- Wants to include a paper in the research-papers collection
Input URL Handling
The user may provide URLs in different formats. Convert them to HTML format for best parsing:
| Input Format | Example | Convert To |
|---|---|---|
| ArXiv HTML | | Use as-is (preferred) |
| ArXiv Abstract | | |
| ArXiv PDF | | |
| AlphaXiv | | Fetch directly |
Workflow
Step 1: Fetch and Parse Paper
- Use
to retrieve the paper content from the HTML URLWebFetch - Extract these key elements:
- Title: Full paper title
- Authors: Author names and affiliations
- Date: Publication/submission date
- Abstract: Paper abstract
- Key Sections: Introduction, methodology, results, conclusion
Step 2: Analyze and Extract Insights
From the paper content, identify:
- 3-5 Key Findings - The most important discoveries or contributions
- Performance Metrics - Specific percentages, improvements, benchmarks
- Use
for improvements<span class="performance-improvement">+24%</span> - Use
for declines<span class="performance-decline">-15%</span> - Use
for neutral metrics<span class="metric">73%</span>
- Use
- Methodology - How the research was conducted
- Benchmarks/Datasets - What was tested and where
- Figure URLs - Extract image URLs from the HTML for inclusion
Step 3: Extract Figure URLs and Determine Sizing
For ArXiv HTML pages, figures are typically at:
- Direct
tags in the HTML content<img> - ArXiv CDN paths
For AlphaXiv papers, use pattern:
https://paper-assets.alphaxiv.org/figures/[paper-id]/img-[N].jpeg
Image Sizing Guidelines
IMPORTANT: Paper figures vary greatly in size. Apply appropriate sizing based on content type:
| Figure Type | Recommended Width | Usage |
|---|---|---|
| Architecture diagrams | | Full width for complex diagrams |
| Most paper figures | | Default - bar charts, results, comparisons |
| Charts with legends | | Pie charts, distribution charts |
| Tables as images | | Comparison tables, results |
| Simple diagrams | | Flowcharts, concept illustrations |
| Very small figures | | Icons, simple symbols |
Apply sizing using inline style on the
tag:<img>
IMPORTANT: Use
width: 100%; max-width: Xpx; — not just max-width alone!
alone won't scale up small imagesmax-width
makes the image fill its containerwidth: 100%
then caps it at a reasonable sizemax-width
<!-- Full width for architecture diagrams --> <div class="figure"> <img src="..." alt="..." style="width: 100%;"> <div class="figure-caption">Figure 1: System Architecture</div> </div> <!-- Default for most figures (RECOMMENDED) --> <div class="figure"> <img src="..." alt="..." style="width: 100%; max-width: 800px;"> <div class="figure-caption">Figure 2: Performance Results</div> </div> <!-- Smaller figures --> <div class="figure"> <img src="..." alt="..." style="width: 100%; max-width: 650px;"> <div class="figure-caption">Figure 3: Concept Diagram</div> </div>
Rules of thumb:
- Academic paper figures are usually high-resolution → use 800px as default
- If the figure has lots of small text/labels → use 100% width
- Only constrain to 500-650px for genuinely small/simple diagrams
- When in doubt, use
— too small is worse than too largemax-width: 800px - Always keep
in the base CSS to prevent overflowmax-width: 100%
Step 4: Generate HTML File
- Read the template from
in this skill directoryTEMPLATE.md - Create file:
/research-papers/[topic]_report.html - Use descriptive filename (e.g.,
)agent_memory_systems_report.html
Required Sections:
1. Navigation header (copy from template) 2. Title with <br> for subtitle 3. Authors div with date 4. Executive Summary (.abstract box) - 2-3 paragraphs with metrics 5. ELI5 box (.eli5-box) - Simple analogy for non-experts 6. Key figures from the paper (.figure with .figure-caption) 7. Part-based content (Part 1:, Part 2:, etc.) 8. Key finding boxes (.key-finding) for important discoveries 9. Methodology boxes (.methodology-box) for methods 10. Tables for comparisons and benchmarks 11. Conclusion box (.conclusion-box) with summary 12. Source box (.source-box) with original paper link 13. Navigation footer (same as header)
Step 5: Update Index
Edit
/research-papers/index.html to add a new table row:
<tr> <td> <a href="[filename].html" target="_blank" class="paper-link"> <strong>[Paper Title]</strong> </a> </td> <td class="description">[150-200 word description with key metrics, methods, and improvements. Focus on specific numbers and practical implications.]</td> </tr>
Add the new entry after the last
<tr> in the <tbody>.
Step 6: Validate
Before completing, verify:
- All required sections present
- Figure URLs are valid and load
- Metrics highlighted with appropriate classes
- Navigation links are correct (
, etc.)../index.html - ELI5 explanation is genuinely simple
- Source box links to original paper
- Index.html entry added with description
Quality Guidelines
Content Standards
- Information Density: Pack maximum useful information per paragraph
- Specific Metrics: Use "24%" not "significant improvement"
- ELI5 Analogies: Use real-world comparisons anyone can understand
- Part Structure: Organize as "Part 1:", "Part 2:", etc.
- Key Findings: Highlight in dedicated boxes
- Honest Limitations: Include what the paper doesn't solve
Description for Index (150-200 words)
The index description should include:
- What the paper introduces (method/framework/analysis)
- Key metrics and improvements (specific percentages)
- Benchmarks or environments tested
- Main implications or breakthroughs
- Academic but accessible tone
File References
- Template: See
in this directory for complete HTML structureTEMPLATE.md - Existing papers: Check
for style reference/research-papers/*.html - Guidelines: See
for detailed documentation/research-papers/CLAUDE.md
Example Usage
User: "Add this paper: https://arxiv.org/html/2512.04123v1"
- Fetch the ArXiv HTML page
- Extract title, authors, abstract, key sections
- Identify 3-5 key findings with metrics
- Find figure URLs in the HTML
- Generate
using template/research-papers/[topic]_report.html - Add entry to
/research-papers/index.html - Report completion with file path