Menu
Parse PDF papers into structured Markdown/JSON using MinerU's precision extraction API.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Beamer LaTeX slide workflow: create, compile, review, and polish academic presentations. Use this skill whenever the user works on Beamer .tex slide decks, or asks to create slides, make a presentation, prepare a lecture, build a talk, or generate Beamer slides from a paper. Covers: creation, editing, compilation, proofreading, visual audit, pedagogical review, TikZ diagrams, figure extraction, and comprehensive quality checks. Trigger on: beamer, slides, lecture, presentation, seminar talk, conference talk, defense slides, tikz, compile latex, proofread slides, slide review, 讨论班, 论文讲解. Do NOT trigger on: powerpoint, pptx, PPT, 做PPT — use the powerpoint-slides skill instead.
Compare reproduced results against paper-reported values. Generate Markdown/JSON/Beamer reports.
Deep analysis of ML source code repositories — AST call graphs, training loop dissection, reproducibility scoring.
Generate implementation code scaffolding from paper descriptions when no source code exists.
Convert LaTeX math formulas from papers into executable PyTorch/NumPy code.
Automate paper reproduction on remote GPU servers via mcp-ssh, using code-analyzer's reports.
| name | paper-parser |
| description | Parse PDF papers into structured Markdown/JSON using MinerU's precision extraction API. |
Parse PDF papers into structured data (Markdown, images, tables, formulas) using MinerU's precision API. This is the third step in the paper reproduction pipeline.
On first use, the user needs to provide their MinerU API token:
set MINERU_TOKEN=<token> (Windows) or export MINERU_TOKEN=<token> (Linux/Mac)The PDF should have been downloaded by paper-downloader, typically at:
workspace/<paper-slug>/paper/paper.pdf
python skills/paper-parser/parse_paper.py workspace/<paper-slug>/paper/paper.pdf -o workspace/<paper-slug>/parsed/
Options:
--model vlm — Highest precision (default, recommended for academic papers)--model pipeline — Faster but less precise--language en — For English papers (default)--language ch — For Chinese papers--timeout 600 — Increase timeout for large papersCheck workspace/<paper-slug>/parsed/ for:
*.md — Full paper in Markdown formatimages/ — Extracted figures and diagramstables/ — Extracted table dataparse_report.json — Parsing log and file listTell the user:
requests library