Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when planning an end-to-end scholarly publishing workflow, including manuscript source-of-truth, submission assets, revision/rebuttal files, camera-ready checks, reproducible build expectations, and publication package structure.
.claude/skills/foryourhealth111-pixel-scholarly-publishing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 86% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 37% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 56% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 75% | 0% |
| case-16 | ✗→✓ | ▲ Improved | 43% | 0% |
当用户说“我要投稿/返修/相机就绪/需要 LaTeX 工程化/写 rebuttal/写 cover letter”时，本 skill 负责把目标拆成可交付的出版资产包：
manuscript/：论文源文件（LaTeX / Word / Markdown 任一作为 source-of-truth）figures/：每张图的源代码/源数据/最终导出要求（PDF/EPS/SVG/TIFF）supplement/：补充材料（方法细节、附录、扩展实验、额外图表）submission/：投稿所需文件（cover letter、graphical abstract、highlights、checklist、打包 zip）revision/：返修资产（rebuttal、diff、逐条回应矩阵）build/：可复现构建产物（PDF、打包 zip、CI 日志）> 目标不是“写一段文字/画一张图”，而是产出能提交、能返修、能复用、能审计的一套文件与规范。
适用场景（中英混合均可）：
投稿、submission、返修、revision、rebuttal、回复审稿意见、camera-ready、prooflatex template、latexmk、bibtex、biber、Overleaf、chktex、latexindentmanuscript as code、reproducible manuscript、submission zip、paper pdf build不适用：
为了稳定落地，至少需要： 1) 目标投向：期刊/会议/出版社（不知道也可以先用“类目”：Nature/IEEE/ACM/NeurIPS/PLOS） 2) 论文类型：研究论文/方法论文/综述/短文/技术报告 3) 交付物：投稿包、返修包、camera-ready 包、可复现构建包、项目主页/视频摘要/海报等传播资产（可多选） 4) 写作来源：是否已有草稿/数据/图？（已有就以“改稿/补齐规范”为主）
在以下三者中选一个做源文件（强烈建议只选一个）：
> 元规则：同一论文不要在多个格式里并行编辑。其它格式只能是“导出物”。
1) 目标投向与模板：记录版式、匿名、页数、图表、引用和补充材料约束 2) 投稿清单：按 pre-submission、submission、revision、camera-ready 四类阶段列出必需文件 3) 明确图的规格：列出每张图的用途、类型（line art / raster / combination）与导出格式（PDF/TIFF）
执行两段式写作：
元规则（顶级期刊通用）：
本 skill 只定义投稿所需的图表交付约束，不在正文内要求调用其它作图专家：
PDF/EPS/SVG（矢量），必要时 TIFF 600dpi（栅格）如果 source-of-truth 是 LaTeX：
如果 source-of-truth 是 Word：
如果需要项目主页、视频摘要或海报，先定义输入来源、目标受众、输出目录和会议/机构限制。本 skill 不负责生成幻灯片，也不在正文内要求调用传播类专家。
fig-01-overview.pdf、fig-02-results.tifffig-02A-...、fig-02B-...kebab-case；避免空格与中文；避免“final_v7_reallyfinal”至少提供：
make pdf / latexmk / quarto render）见：references/case-library.md（按“写作清单/论文工程化/出版包/返修资产”分类）
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 26,991 | 23,901 | -11% | 1 | 1 | 0% | 4,623 | 6,210 | +34% | 0 | 0 | — |
case-02 | fail→fail | 21,439 | 23,667 | +10% | 1 | 1 | 0% | 3,339 | 5,727 | +72% | 0 | 0 | — |
case-03 | fail→pass | 19,036 | 24,790 | +30% | 1 | 1 | 0% | 3,080 | 5,731 | +86% | 0 | 0 | — |
case-04 | fail→pass | 22,329 | 18,823 | -16% | 1 | 1 | 0% | 3,408 | 4,681 | +37% | 0 | 0 | — |
case-05 | pass→pass | 20,746 | 18,526 | -11% | 1 | 1 | 0% | 3,194 | 4,640 | +45% | 0 | 0 | — |
case-06 | pass→pass | 20,673 | 20,723 | +0% | 1 | 1 | 0% | 3,388 | 5,618 | +66% | 0 | 0 | — |
case-07 | fail→pass | 18,166 | 17,734 | -2% | 1 | 1 | 0% | 2,969 | 4,624 | +56% | 0 | 0 | — |
case-08 | pass→fail | 19,462 | 22,781 | +17% | 1 | 1 | 0% | 3,178 | 5,645 | +78% | 0 | 0 | — |
case-09 | fail→fail | 19,263 | 20,213 | +5% | 1 | 1 | 0% | 3,314 | 5,010 | +51% | 0 | 0 | — |
case-10 | fail→fail | 18,684 | 17,649 | -6% | 1 | 1 | 0% | 2,671 | 4,456 | +67% | 0 | 0 | — |
case-11 | fail→pass | 17,274 | 20,472 | +19% | 1 | 1 | 0% | 2,883 | 5,059 | +75% | 0 | 0 | — |
case-12 | pass→pass | 15,753 | 12,381 | -21% | 1 | 1 | 0% | 2,455 | 3,838 | +56% | 0 | 0 | — |
case-13 | pass→pass | 12,512 | 13,862 | +11% | 1 | 1 | 0% | 1,936 | 3,879 | +100% | 0 | 0 | — |
case-14 | pass→pass | 18,836 | 18,297 | -3% | 1 | 1 | 0% | 2,683 | 4,355 | +62% | 0 | 0 | — |
case-15 | pass→pass | 19,696 | 20,776 | +5% | 1 | 1 | 0% | 3,082 | 4,957 | +61% | 0 | 0 | — |
case-16 | fail→pass | 16,677 | 13,444 | -19% | 1 | 1 | 0% | 2,808 | 4,029 | +43% | 0 | 0 | — |
case-17 | fail→pass | 17,663 | 16,926 | -4% | 1 | 1 | 0% | 2,833 | 4,584 | +62% | 0 | 0 | — |
case-18 | pass→pass | 13,300 | 15,468 | +16% | 1 | 1 | 0% | 2,209 | 4,329 | +96% | 0 | 0 | — |
case-19 | pass→pass | 17,444 | 18,836 | +8% | 1 | 1 | 0% | 3,282 | 5,442 | +66% | 0 | 0 | — |
case-20 | pass→pass | 22,888 | 31,542 | +38% | 1 | 1 | 0% | 4,439 | 7,798 | +76% | 0 | 0 | — |
case-21 | pass→pass | 16,315 | 12,251 | -25% | 1 | 1 | 0% | 3,192 | 3,907 | +22% | 0 | 0 | — |
case-22 | fail→fail | 14,534 | 13,343 | -8% | 1 | 1 | 0% | 2,287 | 3,656 | +60% | 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 +23 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.