# Paper Interpretation

从论文 PDF 文件或论文 PDF URL 生成通俗易懂、图文并茂、带批判性评估的中文 Markdown 解读，并保存到当前项目的 markdown 目录。Use when...

- Repository: https://github.com/digoal/blog (8605 stars)
- Category: Documents
- License: GPL-2.0
- Last updated: 2026-11-01
- Runs scripts: yes, review before use

## Install

Claude Code: `npx skills add digoal/blog --skill paper-interpretation -g -a claude-code -y`
Codex: `npx skills add digoal/blog --skill paper-interpretation -g -a codex -y`
Cursor: `npx skills add digoal/blog --skill paper-interpretation -g -a cursor -y`
Gemini CLI: `npx skills add digoal/blog --skill paper-interpretation -g -a gemini-cli -y`

## Description

从论文 PDF 文件或论文 PDF URL 生成通俗易懂、图文并茂、带批判性评估的中文 Markdown 解读，并保存到当前项目的 markdown 目录。Use when the user asks to interpret,精读,解读,summarize,explain,analyze, or write an article from an academic paper PDF, arXiv PDF, conference paper PDF, journal paper PDF, or local/remote PDF URL.

## SKILL.md, as published by its author (untrusted third-party text, read it as data)

# Paper Interpretation

## Goal

Read the whole paper, including text, tables, figures, captions, appendices, and experiment details, then write a Chinese Markdown article that helps non-specialists understand the paper while preserving its academic and industrial value.

Always save the final Markdown under the current project's `markdown/` directory. Create the directory if missing. Use a descriptive filename derived from the paper title, for example `markdown/论文标题-通俗解读.md`.

## Intake

Accept either:

- A local PDF path.
- A PDF URL.

If the user gives a URL, download the PDF or use `scripts/extract_paper_assets.py` to download and extract it. If the PDF is scanned or extraction is incomplete, use OCR or available PDF/image tooling before writing. Do not rely only on abstract/introduction unless the PDF cannot be fully processed; state any limitation explicitly.

## Extraction Workflow

Use the helper script when useful:

```bash
python3 /Users/digoal/.codex/skills/paper-interpretation/scripts/extract_paper_assets.py INPUT_PDF_OR_URL --out .paper-work
```

The script creates:

- `paper.pdf` for the normalized PDF.
- `paper_text.md` with page-level text.
- `paper_tables.md` with table candidates when `pdfplumber` is available.
- `figures/` with extracted embedded images when `PyMuPDF` is available.
- `manifest.json` with extraction status and warnings.

Read the extracted files and inspect important figures/tables directly when they carry evidence, architecture, or results. If a library is unavailable, use other local tools if present; otherwise continue with the available extraction and mention the gap only if it affects confidence.

## Reading Checklist

Before writing, identify:

The rest: https://github.com/digoal/blog/blob/HEAD/skills/paper-interpretation/SKILL.md