Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →11 academic writing skills. Trigger: writing paper sections, structuring arguments, academic prose. Design: section-by-section guides (abstract, intro, methods, discussion) with templates.
.claude/skills/brycewang-stanford-composition-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -37% | 0% |
Select the skill matching the user's need, then read its SKILL.md.
| Skill | Description | |-------|-------------| | abstract-writing-guide | Craft structured research abstracts that maximize clarity and journal acceptance | | discussion-writing-guide | Write effective discussion sections that interpret results and impact | | introduction-writing-guide | Guide to writing effective research paper introductions | | literature-review-writing | Structure and write comprehensive literature reviews for any field | | methods-section-guide | Guide to writing clear and reproducible methodology sections | | ml-paper-writing | Write ML/AI research papers targeting NeurIPS, ICML, and ICLR venues | | research-paper-writer | Guide for writing formal academic papers following IEEE and ACM standards | | response-to-reviewers | Write effective point-by-point responses to peer reviewer comments for revisions | | scientific-writing-guide | Curated tools and techniques for scientific writing beyond LaTeX | | scientific-writing-resources | Curated tools and resources for effective scientific writing | | scientific-writing-wrapper | AI-powered scientific writing workflow from outline to polished draft |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-07 | fail→pass | 9,387 | 2,440 | -74% | 1 | 1 | 0% | 1,485 | 834 | -44% | 0 | 0 | — |
case-01 | fail→fail | 14,412 | 63,396 | +340% | 1 | 1 | 0% | 2,485 | 580 | -77% | 0 | 0 | — |
case-02 | fail→fail | 49,782 | 2,542 | -95% | 1 | 1 | 0% | 2,937 | 564 | -81% | 0 | 0 | — |
case-03 | fail→pass | 7,569 | 2,780 | -63% | 1 | 1 | 0% | 1,176 | 796 | -32% | 0 | 0 | — |
case-04 | fail→pass | 7,301 | 2,800 | -62% | 1 | 1 | 0% | 1,096 | 865 | -21% | 0 | 0 | — |
case-05 | fail→pass | 6,890 | 4,095 | -41% | 1 | 1 | 0% | 1,023 | 996 | -3% | 0 | 0 | — |
case-06 | fail→pass | 12,093 | 7,034 | -42% | 1 | 1 | 0% | 1,689 | 1,056 | -37% | 0 | 0 | — |
case-08 | fail→fail | 14,249 | 3,200 | -78% | 1 | 1 | 0% | 2,328 | 679 | -71% | 0 | 0 | — |
case-09 | fail→pass | 14,071 | 4,816 | -66% | 1 | 1 | 0% | 2,035 | 1,063 | -48% | 0 | 0 | — |
case-10 | fail→fail | 13,470 | 34,100 | +153% | 1 | 1 | 0% | 2,228 | 719 | -68% | 0 | 0 | — |
case-11 | fail→pass | 11,096 | 4,978 | -55% | 1 | 1 | 0% | 1,698 | 946 | -44% | 0 | 0 | — |
case-17 | fail→pass | 13,651 | 35,924 | +163% | 1 | 1 | 0% | 2,143 | 905 | -58% | 0 | 0 | — |
case-12 | fail→pass | 11,128 | 3,292 | -70% | 1 | 1 | 0% | 1,805 | 1,056 | -41% | 0 | 0 | — |
case-13 | fail→pass | 9,348 | 6,961 | -26% | 1 | 1 | 0% | 1,566 | 944 | -40% | 0 | 0 | — |
case-14 | fail→pass | 12,267 | 2,525 | -79% | 1 | 1 | 0% | 1,951 | 806 | -59% | 0 | 0 | — |
case-15 | fail→fail | 15,006 | 19,047 | +27% | 1 | 1 | 0% | 2,130 | 678 | -68% | 0 | 0 | — |
case-16 | fail→fail | 18,401 | 9,436 | -49% | 1 | 1 | 0% | 3,059 | 541 | -82% | 0 | 0 | — |
case-18 | fail→pass | 10,356 | 6,970 | -33% | 1 | 1 | 0% | 1,642 | 952 | -42% | 0 | 0 | — |
case-19 | fail→fail | 6,848 | 6,303 | -8% | 1 | 1 | 0% | 1,008 | 956 | -5% | 0 | 0 | — |
case-20 | pass→fail | 20,036 | 4,987 | -75% | 1 | 1 | 0% | 3,273 | 803 | -75% | 0 | 0 | — |
case-21 | pass→pass | 35,822 | 13,807 | -61% | 1 | 1 | 0% | 2,792 | 2,469 | -12% | 0 | 0 | — |
case-22 | pass→pass | 15,955 | 16,060 | +1% | 1 | 1 | 0% | 2,271 | 2,796 | +23% | 0 | 0 | — |
case-23 | fail→pass | 17,523 | 4,800 | -73% | 1 | 1 | 0% | 2,046 | 1,139 | -44% | 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. 23 cases were attempted, and 15 counted toward the lift figure. The other 8 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 +52 percentage points is the difference between those two pass rates over the 15 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.