nature-skills is an open-source project by Yuan Yizhe, a medical AI PhD student at Shanghai Jiao Tong University, that packages the full workflow of writing Nature-level papers into 9 installable AI skills. The pipeline covers language polishing (nature-polishing), argument-driven drafting (nature-writing), publication-grade figures with matplotlib/ggplot2 (nature-figure), citation management (nature-citation), data availability statements (nature-data), bilingual paper reading (nature-reader), point-by-point peer-review responses (nature-response), paper-to-Chinese-PPT conversion (nature-paper2ppt), and parallel PubMed/CrossRef/arXiv search (nature-academic-search). Key design principles include a six-layer diagnostic before sentence-level polishing, 30-word sentence limits, calibrated hedging, figure contracts written before plotting, evidence-first writing order, and an explicit red/yellow/green traffic-light policy for what AI should never do. Only two skills are currently Stable; the rest are Beta/Draft. The project installs via Claude Code or Codex Desktop plugin marketplaces and is extensible with planned skills for statistics, methods, cover letters, and reviews.
nature-skills Deep Dive: How a Shanghai Jiao Tong PhD Student Turned Nature Paper Writing into Replicable AI Skills
Summary
nature-skills is an open-source project by Yuan Yizhe, a medical AI PhD student at Shanghai Jiao Tong University, that packages the full workflow of writing Nature-level papers into 9 installable AI skills. The pipeline covers language polishing (nature-polishing), argument-driven drafting (nature-writing), publication-grade figures with matplotlib/ggplot2 (nature-figure), citation management (nature-citation), data availability statements (nature-data), bilingual paper reading (nature-reader), point-by-point peer-review responses (nature-response), paper-to-Chinese-PPT conversion (nature-paper2ppt), and parallel PubMed/CrossRef/arXiv search (nature-academic-search). Key design principles include a six-layer diagnostic before sentence-level polishing, 30-word sentence limits, calibrated hedging, figure contracts written before plotting, evidence-first writing order, and an explicit red/yellow/green traffic-light policy for what AI should never do. Only two skills are currently Stable; the rest are Beta/Draft. The project installs via Claude Code or Codex Desktop plugin marketplaces and is extensible with planned skills for statistics, methods, cover letters, and reviews.
This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177980441