Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →9 paper discovery skills. Trigger: finding new relevant papers, tracking citations, staying current. Design: automated monitoring, recommendation engines, and alert setup guides.
.claude/skills/brycewang-stanford-discovery-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | -59% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -68% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -9% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -57% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -40% | 0% |
Select the skill matching the user's need, then read its SKILL.md.
| Skill | Description | |-------|-------------| | citation-alert-guide | Set up citation alerts and track new papers citing key references | | conference-proceedings-guide | Find, access, and cite conference papers and proceedings effectively | | literature-mapping-guide | Visual literature mapping and connected papers exploration | | paper-recommendation-guide | Systematic paper recommendation and discovery using multiple methods | | papers-we-love-guide | Community-curated directory of influential CS research papers | | rss-paper-feeds | Set up RSS feeds and alerts to track new publications in your research area | | semantic-paper-radar | Semantic literature discovery and synthesis using embeddings | | semantic-scholar-recs-guide | Paper discovery via recommendation APIs (OpenAlex, CrossRef citation networks) | | zotero-arxiv-daily-guide | Guide to Zotero arXiv Daily for personalized daily paper recommendations |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,919 | 65,906 | +342% | 1 | 1 | 0% | 2,792 | 852 | -69% | 0 | 0 | — |
case-02 | fail→fail | 15,033 | 34,984 | +133% | 1 | 1 | 0% | 2,722 | 707 | -74% | 0 | 0 | — |
case-03 | fail→pass | 17,690 | 3,147 | -82% | 1 | 1 | 0% | 2,565 | 1,056 | -59% | 0 | 0 | — |
case-22 | pass→pass | 17,676 | 11,759 | -33% | 1 | 1 | 0% | 2,564 | 2,384 | -7% | 0 | 0 | — |
case-04 | fail→pass | 42,634 | 2,282 | -95% | 1 | 1 | 0% | 2,477 | 790 | -68% | 0 | 0 | — |
case-05 | fail→pass | 4,739 | 3,018 | -36% | 1 | 1 | 0% | 890 | 812 | -9% | 0 | 0 | — |
case-06 | fail→pass | 11,261 | 2,179 | -81% | 1 | 1 | 0% | 1,857 | 799 | -57% | 0 | 0 | — |
case-07 | fail→pass | 8,333 | 2,422 | -71% | 1 | 1 | 0% | 1,337 | 800 | -40% | 0 | 0 | — |
case-08 | fail→pass | 11,935 | 2,681 | -78% | 1 | 1 | 0% | 2,189 | 906 | -59% | 0 | 0 | — |
case-09 | fail→pass | 13,470 | 1,990 | -85% | 1 | 1 | 0% | 2,364 | 760 | -68% | 0 | 0 | — |
case-10 | pass→pass | 15,514 | 12,250 | -21% | 1 | 1 | 0% | 2,374 | 2,418 | +2% | 0 | 0 | — |
case-11 | pass→pass | 14,799 | 13,303 | -10% | 1 | 1 | 0% | 1,947 | 2,571 | +32% | 0 | 0 | — |
case-12 | pass→pass | 12,013 | 11,063 | -8% | 1 | 1 | 0% | 2,389 | 2,135 | -11% | 0 | 0 | — |
case-13 | fail→pass | 7,238 | 2,123 | -71% | 1 | 1 | 0% | 1,164 | 787 | -32% | 0 | 0 | — |
case-14 | fail→pass | 12,130 | 3,674 | -70% | 1 | 1 | 0% | 2,196 | 903 | -59% | 0 | 0 | — |
case-15 | fail→pass | 2,370 | 3,080 | +30% | 1 | 1 | 0% | 387 | 850 | +120% | 0 | 0 | — |
case-16 | fail→pass | 8,226 | 1,873 | -77% | 1 | 1 | 0% | 1,468 | 736 | -50% | 0 | 0 | — |
case-17 | fail→pass | 8,305 | 3,502 | -58% | 1 | 1 | 0% | 1,311 | 940 | -28% | 0 | 0 | — |
case-18 | fail→pass | 12,976 | 2,774 | -79% | 1 | 1 | 0% | 1,940 | 831 | -57% | 0 | 0 | — |
case-19 | fail→pass | 7,208 | 4,092 | -43% | 1 | 1 | 0% | 1,343 | 1,013 | -25% | 0 | 0 | — |
case-20 | fail→pass | 15,199 | 3,185 | -79% | 1 | 1 | 0% | 2,340 | 972 | -58% | 0 | 0 | — |
case-21 | fail→pass | 7,454 | 2,759 | -63% | 1 | 1 | 0% | 1,161 | 805 | -31% | 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 +73 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.