aba122/researchpaper-analyzer

🔬 > **Just paste the paper link, and in-depth technical articles will be automatically generated. ** PDF auto-download, MinerU high-precision parsing, Claude in-depth analysis—images, tables, formulas all included, directly stored in your Obsidian knowledge base.

🔗 > ** Includes GitHub repository for code-level interpretation. ** Precisely pinpoints the file name and line number corresponding to each core innovation, breaks down the model architecture and training process, and accompanies real code snippets.

• 💡 One-click access from URLs to structured knowledge. 🌱*

---

🎯 Core Features

Paste the arxiv link → configuration options → get an in-depth technical article including:

• MinerU parsed images and experiment tables, fully embedded in the article
• LaTeX formulas KaTeX real-time rendering
• Three writing styles: Academic / Narrative / Concise Writing
• Code-level analysis (optional): Model architecture, training flow, core innovations—down to files and line numbers
• Auto-deposit into Obsidian, carry the full YAML frontmatter (Title / Author / arxiv / Tag)
• Agent Dialogue: After analysis is complete, continue asking questions or requesting modifications

---

🏆 Core Advantages

1. 📚 Deeply integrated with Obsidian to build a personal thesis knowledge base

All analysis results are automatically saved to the vault, carrying the complete YAML frontmatter, allowing direct browsing, searching, and two-way linking within Obsidian. The paper library supports tag filtering, and literature management is well organized.

2. 🖼️ A richly illustrated in-depth technical article, not just a simple summary

MinerU precisely extracts images, tables, and formulas from papers, fully embedding them into the generated article, preserving the visual evidence of the original paper and presenting in-depth, structured technical interpretations.

3. 🔍 Integration with GitHub repository to pinpoint code-level innovations

After providing the repository link, Claude will:

• Filter key source code files (50k token budget)
• Precisely pinpoint the code implementation for each innovation point — filename + line number range
• Explain the model architecture using real __init__ and forward code
• Sort out training loops, loss functions, and data processing workflows

Freedom of Choice of Analytical Dimensions: Model Architecture · Training Process · Main Innovations

---

📸 Interface Preview

Search and Configuration

Enter the arxiv link or title, choose the writing style, and start formula explanation and code analysis:

  

Analysis Report

The left side displays the progress of each stage in real time, the right side renders a complete technical article with images, tables, and formulas, and the upper right corner allows you to talk to the Agent:

In-depth Code Analysis

Model architecture, training paradigms, major innovations—each comes with precise documentation + line number references:

---

🧩 Depend on Skill: paper-analyzer