Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →10 research automation skills. Trigger: automating experiments, tracking results, reproducible pipelines. Design: ML experiment management, workflow orchestration, and lab automation tools.
.claude/skills/brycewang-stanford-automation-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-12 | ✗→✓ | ▲ Improved | -71% | 0% |
| case-14 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-21 | ✓→✗ | ▼ Worse | -65% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -67% | 0% |
| case-23 | ✓→✓ | ＝ Same ✓ | 36% | 0% |
Select the skill matching the user's need, then read its SKILL.md.
| Skill | Description | |-------|-------------| | ai-scientist-v2-guide | Automated scientific discovery via agentic tree search by Sakana AI | | aim-experiment-guide | Track and compare research experiments with Aim experiment tracker | | claude-academic-workflow-guide | Claude Code template for LaTeX, Beamer, and R research workflows | | data-collection-automation | Automate survey deployment, data collection, and pipeline management | | datagen-research-guide | AI-driven multi-agent research assistant for end-to-end studies | | kedro-pipeline-guide | Build reproducible data science pipelines with Kedro for research projects | | mle-agent-guide | Intelligent companion for ML engineering with arXiv integration | | paper-to-agent-guide | Transform research papers into interactive AI agents for exploration | | rd-agent-guide | Microsoft AI-driven R&D agent for automated data and model development | | research-workflow-automation | Automate repetitive research tasks with pipelines, schedulers, and scripting |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 18,853 | 5,027 | -73% | 1 | 1 | 0% | 3,414 | 713 | -79% | 0 | 0 | — |
case-02 | fail→fail | 22,120 | 3,859 | -83% | 1 | 1 | 0% | 3,673 | 549 | -85% | 0 | 0 | — |
case-03 | fail→fail | 20,878 | 4,785 | -77% | 1 | 1 | 0% | 3,868 | 681 | -82% | 0 | 0 | — |
case-04 | fail→fail | 13,435 | 2,622 | -80% | 1 | 1 | 0% | 2,301 | 673 | -71% | 0 | 0 | — |
case-05 | fail→fail | 17,352 | 3,595 | -79% | 1 | 1 | 0% | 3,108 | 561 | -82% | 0 | 0 | — |
case-06 | fail→fail | 15,205 | 3,093 | -80% | 1 | 1 | 0% | 2,814 | 635 | -77% | 0 | 0 | — |
case-07 | fail→fail | 28,741 | 6,064 | -79% | 1 | 1 | 0% | 5,503 | 784 | -86% | 0 | 0 | — |
case-08 | fail→fail | 20,375 | 3,944 | -81% | 1 | 1 | 0% | 3,458 | 755 | -78% | 0 | 0 | — |
case-09 | fail→fail | 18,699 | 3,346 | -82% | 1 | 1 | 0% | 3,168 | 661 | -79% | 0 | 0 | — |
case-10 | fail→fail | 14,818 | 4,873 | -67% | 1 | 1 | 0% | 2,396 | 954 | -60% | 0 | 0 | — |
case-11 | fail→fail | 19,428 | 4,207 | -78% | 1 | 1 | 0% | 3,900 | 597 | -85% | 0 | 0 | — |
case-12 | fail→pass | 16,468 | 2,928 | -82% | 1 | 1 | 0% | 3,018 | 861 | -71% | 0 | 0 | — |
case-13 | fail→fail | 16,273 | 5,167 | -68% | 1 | 1 | 0% | 2,721 | 808 | -70% | 0 | 0 | — |
case-19 | fail→fail | 42,683 | 5,795 | -86% | 1 | 1 | 0% | 3,336 | 802 | -76% | 0 | 0 | — |
case-14 | fail→pass | 11,349 | 3,465 | -69% | 1 | 1 | 0% | 1,861 | 939 | -50% | 0 | 0 | — |
case-15 | fail→fail | 20,213 | 4,836 | -76% | 1 | 1 | 0% | 3,850 | 807 | -79% | 0 | 0 | — |
case-16 | fail→fail | 19,118 | 3,549 | -81% | 1 | 1 | 0% | 3,036 | 650 | -79% | 0 | 0 | — |
case-17 | fail→fail | 20,430 | 3,672 | -82% | 1 | 1 | 0% | 3,443 | 631 | -82% | 0 | 0 | — |
case-18 | fail→fail | 11,105 | 13,609 | +23% | 1 | 1 | 0% | 1,946 | 591 | -70% | 0 | 0 | — |
case-20 | fail→fail | 16,501 | 4,669 | -72% | 1 | 1 | 0% | 2,722 | 686 | -75% | 0 | 0 | — |
case-21 | pass→fail | 16,379 | 5,911 | -64% | 1 | 1 | 0% | 2,557 | 897 | -65% | 0 | 0 | — |
case-22 | pass→fail | 15,373 | 4,784 | -69% | 1 | 1 | 0% | 2,353 | 773 | -67% | 0 | 0 | — |
case-23 | pass→pass | 20,192 | 23,635 | +17% | 1 | 1 | 0% | 3,434 | 4,655 | +36% | 0 | 0 | — |
case-24 | pass→pass | 14,857 | 15,500 | +4% | 1 | 1 | 0% | 2,262 | 2,727 | +21% | 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. 24 cases were attempted, and 4 counted toward the lift figure. The other 20 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 0 percentage points is the difference between those two pass rates over the 4 comparable cases. 2 cases got worse with the skill loaded, and they are 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.