Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Beautiful LaTeX template for working papers and technical reports
.claude/skills/brycewang-stanford-elegant-paper-template/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-14 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | -5% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 29% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 1% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 64% | 0% |
ElegantPaper is a beautifully designed LaTeX template from the ElegantLaTeX project, tailored for working papers, technical reports, and preprints. It features clean typography, optional color themes (green, cyan, blue, black), bilingual support (English/Chinese), and a minimal yet professional appearance. Part of the ElegantLaTeX series (ElegantPaper, ElegantBook, ElegantNote).
bash# Download template git clone https://github.com/ElegantLaTeX/ElegantPaper.git # Compile with XeLaTeX (for Chinese) or PDFLaTeX (English only) xelatex elegantpaper-en && bibtex elegantpaper-en && xelatex elegantpaper-en
latex\documentclass[lang=en]{elegantpaper} \title{Your Working Paper Title} \author{Author Name\thanks{Affiliation, email@university.edu}} \date{\today} \begin{document} \maketitle \begin{abstract} A concise summary of your working paper. \keywords{keyword1, keyword2, keyword3} \end{abstract} \section{Introduction} Your introduction text here. \section{Model} Mathematical model or methodology. \begin{theorem}\label{thm:main} For all $x \in \mathbb{R}^n$, if $f$ is convex, then... \end{theorem} \begin{proof} The proof follows from... \end{proof} \section{Results} Empirical or theoretical results. \bibliography{references} \end{document}
latex% Language: en (English) or cn (Chinese) \documentclass[lang=en]{elegantpaper} \documentclass[lang=cn]{elegantpaper} % Color theme \documentclass[color=green]{elegantpaper} % Default green \documentclass[color=cyan]{elegantpaper} \documentclass[color=blue]{elegantpaper} \documentclass[color=black]{elegantpaper} % Formal/print % Math font \documentclass[math=cm]{elegantpaper} % Computer Modern \documentclass[math=newtx]{elegantpaper} % Times-like % Citation style \documentclass[cite=authoryear]{elegantpaper} % (Author, Year) \documentclass[cite=numbers]{elegantpaper} % [1]
latex% Theorem family (auto-numbered, colored) \begin{theorem}...\end{theorem} \begin{lemma}...\end{lemma} \begin{proposition}...\end{proposition} \begin{corollary}...\end{corollary} % Definition and examples \begin{definition}...\end{definition} \begin{example}...\end{example} \begin{remark}...\end{remark} % Proof \begin{proof}...\end{proof}
latex\documentclass[lang=cn]{elegantpaper} \title{基于深度学习的自然语言处理研究进展} \author{张三\thanks{北京大学, zhangsan@pku.edu.cn}} \begin{document} \maketitle \begin{abstract} 本文综述了深度学习在自然语言处理中的最新进展。 \keywords{深度学习, 自然语言处理, 注意力机制} \end{abstract} \section{引言} 近年来... \end{document}
| Template | Purpose | |----------|---------| | ElegantPaper | Working papers, short reports | | ElegantBook | Books, long-form documents | | ElegantNote | Lecture notes, course materials |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 7,981 | 10,933 | +37% | 1 | 1 | 0% | 1,417 | 1,344 | -5% | 0 | 0 | — |
case-02 | pass→pass | 9,812 | 6,800 | -31% | 1 | 1 | 0% | 1,658 | 2,141 | +29% | 0 | 0 | — |
case-03 | pass→pass | 10,431 | 4,216 | -60% | 1 | 1 | 0% | 1,515 | 1,532 | +1% | 0 | 0 | — |
case-04 | pass→pass | 4,449 | 2,009 | -55% | 1 | 1 | 0% | 766 | 1,260 | +64% | 0 | 0 | — |
case-05 | pass→pass | 5,408 | 2,752 | -49% | 1 | 1 | 0% | 805 | 1,333 | +66% | 0 | 0 | — |
case-06 | pass→pass | 4,788 | 2,570 | -46% | 1 | 1 | 0% | 789 | 1,339 | +70% | 0 | 0 | — |
case-07 | pass→pass | 14,691 | 2,835 | -81% | 1 | 1 | 0% | 1,503 | 1,374 | -9% | 0 | 0 | — |
case-08 | pass→pass | 6,754 | 2,785 | -59% | 1 | 1 | 0% | 1,176 | 1,347 | +15% | 0 | 0 | — |
case-09 | pass→pass | 10,687 | 3,476 | -67% | 1 | 1 | 0% | 1,949 | 1,443 | -26% | 0 | 0 | — |
case-10 | pass→pass | 9,419 | 2,074 | -78% | 1 | 1 | 0% | 1,194 | 1,280 | +7% | 0 | 0 | — |
case-11 | pass→pass | 6,686 | 3,195 | -52% | 1 | 1 | 0% | 1,114 | 1,413 | +27% | 0 | 0 | — |
case-12 | pass→pass | 7,438 | 2,235 | -70% | 1 | 1 | 0% | 1,085 | 1,268 | +17% | 0 | 0 | — |
case-13 | pass→pass | 5,657 | 2,375 | -58% | 1 | 1 | 0% | 835 | 1,318 | +58% | 0 | 0 | — |
case-14 | fail→pass | 6,407 | 3,061 | -52% | 1 | 1 | 0% | 1,170 | 1,535 | +31% | 0 | 0 | — |
case-15 | pass→pass | 4,862 | 3,405 | -30% | 1 | 1 | 0% | 871 | 1,338 | +54% | 0 | 0 | — |
case-16 | pass→pass | 4,858 | 3,064 | -37% | 1 | 1 | 0% | 683 | 1,228 | +80% | 0 | 0 | — |
case-17 | pass→pass | 7,562 | 5,844 | -23% | 1 | 1 | 0% | 1,309 | 1,760 | +34% | 0 | 0 | — |
case-18 | pass→pass | 5,470 | 2,926 | -47% | 1 | 1 | 0% | 938 | 1,383 | +47% | 0 | 0 | — |
case-19 | pass→pass | 7,332 | 5,151 | -30% | 1 | 1 | 0% | 1,261 | 1,746 | +38% | 0 | 0 | — |
case-20 | pass→pass | 31,918 | 7,671 | -76% | 1 | 1 | 0% | 1,719 | 2,275 | +32% | 0 | 0 | — |
case-21 | pass→pass | 12,522 | 13,896 | +11% | 1 | 1 | 0% | 2,116 | 3,305 | +56% | 0 | 0 | — |
case-22 | pass→pass | 11,935 | 8,792 | -26% | 1 | 1 | 0% | 1,980 | 2,440 | +23% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.