Install any skill in seconds. Free to start, no credit card required.
Get Started Free →LaTeX thesis template supporting multiple universities and formats
.claude/skills/brycewang-stanford-novathesis-guide/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 70% | 0% |
| case-04 | ✓→✗ | ▼ Worse | 84% | 0% |
A skill for using and customizing the NovaThesis LaTeX thesis template, which supports formatting requirements from multiple universities worldwide. Based on the novathesis project (955 stars), this skill helps graduate students and researchers produce professionally typeset theses and dissertations that comply with institutional formatting mandates.
Writing a thesis is a monumental academic achievement, but formatting it to meet university requirements can be frustratingly time-consuming. NovaThesis eliminates this burden by providing a single LaTeX template that adapts to the formatting requirements of dozens of universities. Students focus on their content while the template handles margins, fonts, title pages, chapter headings, bibliographies, and all the other formatting details that institutions demand.
This skill guides the agent through template selection, customization, content structuring, and troubleshooting, making the thesis writing process smoother for LaTeX beginners and experienced users alike.
Choosing the Right Configuration
Initial Setup
template.tex or thesis.tex)Directory Structure
Chapters/ - individual chapter files, one per chapterAppendices/ - supplementary material filesFigures/ - all figure files organized by chapterBibliography/ - BibTeX database filesConfig/ - template configuration and customization files\include commandsFront Matter
Main Chapters
Back Matter
\index{} commands throughout the textTypography
Page Layout
Bibliography Styles
Adding a New Chapter
\include{Chapters/newchapter} command in the main file\label and \refManaging Figures
\includegraphics with relative paths from the main filesubcaption package (included in the template)Handling References
\cite{}, \citep{}, or \citet{} depending on the required citation styleBilingual Thesis
Common Compilation Issues
Formatting Overrides
This skill supports the Research-Claw thesis writing workflow:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 32,393 | 19,891 | -39% | 1 | 1 | 0% | 3,588 | 4,633 | +29% | 0 | 0 | — |
case-02 | fail→pass | 15,827 | 12,493 | -21% | 1 | 1 | 0% | 2,852 | 3,748 | +31% | 0 | 0 | — |
case-03 | pass→pass | 28,868 | 14,205 | -51% | 1 | 1 | 0% | 2,665 | 3,807 | +43% | 0 | 0 | — |
case-04 | pass→fail | 10,304 | 14,218 | +38% | 1 | 1 | 0% | 1,983 | 3,645 | +84% | 0 | 0 | — |
case-05 | pass→pass | 11,742 | 8,354 | -29% | 1 | 1 | 0% | 1,315 | 2,837 | +116% | 0 | 0 | — |
case-06 | pass→pass | 15,117 | 8,154 | -46% | 1 | 1 | 0% | 2,556 | 2,788 | +9% | 0 | 0 | — |
case-07 | fail→pass | 13,480 | 11,466 | -15% | 1 | 1 | 0% | 2,171 | 3,240 | +49% | 0 | 0 | — |
case-08 | pass→pass | 11,810 | 9,624 | -19% | 1 | 1 | 0% | 2,012 | 2,963 | +47% | 0 | 0 | — |
case-09 | pass→pass | 16,097 | 11,320 | -30% | 1 | 1 | 0% | 2,349 | 3,181 | +35% | 0 | 0 | — |
case-10 | fail→pass | 14,600 | 20,256 | +39% | 1 | 1 | 0% | 2,486 | 3,612 | +45% | 0 | 0 | — |
case-11 | pass→pass | 6,610 | 5,351 | -19% | 1 | 1 | 0% | 1,229 | 2,322 | +89% | 0 | 0 | — |
case-12 | pass→pass | 9,131 | 9,484 | +4% | 1 | 1 | 0% | 1,679 | 3,057 | +82% | 0 | 0 | — |
case-13 | pass→pass | 11,694 | 12,890 | +10% | 1 | 1 | 0% | 1,705 | 3,553 | +108% | 0 | 0 | — |
case-14 | pass→pass | 12,927 | 20,345 | +57% | 1 | 1 | 0% | 1,879 | 3,260 | +73% | 0 | 0 | — |
case-15 | pass→pass | 16,696 | 13,429 | -20% | 1 | 1 | 0% | 2,426 | 3,235 | +33% | 0 | 0 | — |
case-16 | pass→pass | 13,730 | 13,653 | -1% | 1 | 1 | 0% | 2,133 | 3,354 | +57% | 0 | 0 | — |
case-17 | pass→pass | 8,912 | 6,072 | -32% | 1 | 1 | 0% | 1,324 | 2,543 | +92% | 0 | 0 | — |
case-18 | fail→pass | 13,617 | 14,429 | +6% | 1 | 1 | 0% | 1,959 | 3,337 | +70% | 0 | 0 | — |
case-19 | pass→pass | 11,892 | 11,499 | -3% | 1 | 1 | 0% | 1,767 | 3,335 | +89% | 0 | 0 | — |
case-20 | fail→fail | 10,169 | 9,838 | -3% | 1 | 1 | 0% | 1,529 | 2,957 | +93% | 0 | 0 | — |
case-21 | pass→pass | 11,503 | 11,672 | +1% | 1 | 1 | 0% | 2,003 | 3,492 | +74% | 0 | 0 | — |
case-22 | pass→pass | 17,708 | 15,474 | -13% | 1 | 1 | 0% | 2,391 | 3,744 | +57% | 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 +14 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.