Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when building a LaTeX manuscript submission pipeline with templates, latexmk, BibTeX/Biber, chktex, latexindent, CI PDF builds, compile debugging, and submission zip packaging.
.claude/skills/foryourhealth111-pixel-latex-submission-pipeline/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | 22% | 0% |
| case-07 | ✓→✓ | = Same ✓ | 14% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 10% | 0% |
| case-03 | ✓→✓ | = Same ✓ | 26% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 28% | 0% |
当用户需要:
先记录目标 venue 的官方模板或格式要求:
推荐统一入口(任选其一):
latexmk -pdf -interaction=nonstopmode -halt-on-errorscripts/build.ps1 / scripts/build.sh元规则:
.bib 是 source-of-truth对投稿系统最稳的组合:
PDF(matplotlib 导出)TIFF 600dpi(显微/照片/heatmap 视情况)chktex(常见 LaTeX 坑)latexindent(减少 diff 噪声)模板:templates/github-actions-latex.yml(参考 xu-cheng/latex-action)
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-13 | fail→pass | 16,447 | 17,993 | +9% | 1 | 1 | 0% | 2,860 | 3,502 | +22% | 0 | 0 | — |
case-07 | pass→pass | 12,505 | 11,467 | -8% | 1 | 1 | 0% | 2,138 | 2,448 | +14% | 0 | 0 | — |
case-12 | fail→fail | 12,939 | 8,771 | -32% | 1 | 1 | 0% | 2,053 | 1,953 | -5% | 0 | 0 | — |
case-01 | fail→fail | 18,071 | 33,655 | +86% | 1 | 1 | 0% | 3,528 | 1,844 | -48% | 0 | 0 | — |
case-02 | pass→pass | 8,547 | 7,623 | -11% | 1 | 1 | 0% | 1,694 | 1,865 | +10% | 0 | 0 | — |
case-03 | pass→pass | 12,118 | 11,807 | -3% | 1 | 1 | 0% | 2,072 | 2,613 | +26% | 0 | 0 | — |
case-04 | pass→pass | 13,709 | 14,818 | +8% | 1 | 1 | 0% | 2,282 | 2,932 | +28% | 0 | 0 | — |
case-05 | pass→pass | 12,712 | 11,711 | -8% | 1 | 1 | 0% | 1,863 | 2,321 | +25% | 0 | 0 | — |
case-06 | pass→pass | 11,753 | 12,848 | +9% | 1 | 1 | 0% | 2,120 | 2,768 | +31% | 0 | 0 | — |
case-08 | pass→pass | 7,733 | 6,727 | -13% | 1 | 1 | 0% | 1,515 | 1,617 | +7% | 0 | 0 | — |
case-09 | fail→fail | 12,204 | 9,764 | -20% | 1 | 1 | 0% | 2,226 | 2,205 | -1% | 0 | 0 | — |
case-10 | pass→pass | 13,905 | 13,859 | -0% | 1 | 1 | 0% | 2,412 | 2,761 | +14% | 0 | 0 | — |
case-11 | pass→pass | 9,576 | 10,237 | +7% | 1 | 1 | 0% | 1,477 | 2,164 | +47% | 0 | 0 | — |
case-14 | pass→pass | 11,107 | 11,822 | +6% | 1 | 1 | 0% | 2,037 | 2,649 | +30% | 0 | 0 | — |
case-15 | pass→pass | 9,776 | 10,468 | +7% | 1 | 1 | 0% | 1,709 | 2,367 | +39% | 0 | 0 | — |
case-16 | fail→fail | 15,355 | 14,185 | -8% | 1 | 1 | 0% | 2,590 | 3,065 | +18% | 0 | 0 | — |
case-17 | pass→pass | 9,218 | 9,435 | +2% | 1 | 1 | 0% | 1,703 | 2,250 | +32% | 0 | 0 | — |
case-18 | pass→pass | 8,146 | 4,407 | -46% | 1 | 1 | 0% | 1,446 | 1,160 | -20% | 0 | 0 | — |
case-19 | pass→pass | 11,754 | 12,626 | +7% | 1 | 1 | 0% | 1,930 | 2,646 | +37% | 0 | 0 | — |
case-20 | pass→pass | 13,081 | 16,346 | +25% | 1 | 1 | 0% | 2,202 | 3,261 | +48% | 0 | 0 | — |
case-21 | pass→pass | 14,337 | 14,350 | +0% | 1 | 1 | 0% | 2,426 | 3,079 | +27% | 0 | 0 | — |
case-22 | pass→pass | 12,803 | 13,141 | +3% | 1 | 1 | 0% | 2,155 | 2,696 | +25% | 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, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +5 percentage points is the difference between those two pass rates over the 21 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.