claude-paper

Features

• Automatic PDF parsing - Extract title, authors, abstract, and full content
• Smart content truncation - Handles large papers (50k char limit) intelligently
• Code repository detection - Automatically finds GitHub, arXiv, CodeOcean links
• Adaptive learning materials - Generates README, summary, insights, Q&A based on paper complexity
• Code demonstrations - Clean implementations with Jupyter notebooks and original code integration
• Interactive web viewer - Nuxt.js interface with math equation support (KaTeX)
• Intelligent assessment - Difficulty levels and paper type detection for adaptive content generation

---

Quick Start

Installation

Install from the Claude Code marketplace:

# Add the marketplace
/plugin marketplace add alaliqing/claude-paper

# Install the plugin
/plugin install claude-paper

# Restart Claude Code for the plugin to take effect

That's it! The plugin will automatically:

• Install all dependencies (pdf-parse for PDF processing)
• Create the papers directory at ~/claude-papers/
• Initialize the search index
• Install web viewer dependencies

System Requirements

• Node.js: 18.0.0 or higher
• npm: Comes with Node.js
• Claude Code: Latest version with plugin support
• poppler-utils: For PDF image extraction (install via system package manager)
  • macOS: brew install poppler
  • Ubuntu/Debian: sudo apt-get install poppler-utils
  • Arch Linux: sudo pacman -S poppler

---

Usage

Study a Research Paper

Simply talk to Claude Code to study a paper:

Help me study the paper at ~/Downloads/attention-is-all-you-need.pdf

You can also use URLs:

# Direct PDF URL
Help me study the paper at https://arxiv.org/pdf/1706.03762.pdf

# arXiv abstract URL (automatically converted to PDF)
Help me study the paper at https://arxiv.org/abs/1706.03762

Claude will automatically trigger the study workflow and:

1. Parse the PDF and extract metadata
2. Analyze paper complexity and type
3. Generate adaptive learning materials
4. Create code demonstrations (if applicable)
5. Extract and include original code (if available)
6. Extract key figures and images
7. Update the global search index
8. Launch the web viewer automatically

Launch Web Viewer

/claude-paper:webui

Opens the interactive web interface at http://localhost:5815 where you can:

• Browse all studied papers
• View generated materials with math rendering
• Access code demonstrations and notebooks
• Search through your paper library

---

Paper Storage Structure

Papers are organized in ~/claude-papers/papers/{paper-slug}/:

~/claude-papers/
├── papers/
│   └── {paper-slug}/
│       ├── paper.pdf                     # Original PDF file
│       ├── meta.json                     # Paper metadata (title, authors, etc.)
│       ├── README.md                     # Quick navigation and overview
│       ├── summary.md                    # Detailed summary
│       ├── insights.md                   # Key insights (most important!)
│       ├── method.md                     # Methodology (if complex)
│       ├── mental-model.md              # Paper categorization (if needed)
│       ├── reflection.md                # Future directions (if needed)
│       ├── qa.md                         # Learning questions
│       ├── index.html                    # Interactive HTML explorer
│       ├── images/                       # Extracted figures and tables
│       │   ├── fig1.png
│       │   └── fig2.png
│       └── code/                         # Code demonstrations
│           ├── core-demo.py              # Clean reference implementation
│           └── concept-demo.ipynb        # Interactive Jupyter notebook
│
└── index.json                           # Global search index

---

Architecture

Plugin Structure