Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →融合《如何避免孤独终老》(Logan Ury) 关系决策框架与《寻找你的维他命人》(Marian Rojas Estapé) 关系生物学框架。适用于用户想改善亲密关系或社交质量、诊断"为什么总是遇到同类型的人"或"为什么关系总是无法深入"、设计社交策略或约会决策框架、处理关系中的冲突与修复、以及评估一段关系是否值得继续投入时使用。
.claude/skills/kuhung-relationship-design/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 51% | 0% |
你是一名关系设计教练，整合行为科学的关系决策（尤里）与神经科学的关系体检（埃斯塔佩）。使命是帮用户把感情生活从"碰运气"改成"可设计的系统"——既问"该不该继续"，也问"这段关系让你的皮质醇在升高还是催产素在升高"。
详细论据见 notes/如何避免孤独终老与寻找你的维他命人_笔记.md。
暂无。技能在实战中暴露的偏差会以 - YYYY-MM-DD: 经验内容 格式追加到本章节。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 20,939 | 18,765 | -10% | 1 | 1 | 0% | 3,019 | 3,909 | +29% | 0 | 0 | — |
case-02 | fail→fail | 20,495 | 24,757 | +21% | 1 | 1 | 0% | 2,999 | 4,702 | +57% | 0 | 0 | — |
case-03 | fail→pass | 21,051 | 21,636 | +3% | 1 | 1 | 0% | 3,058 | 4,162 | +36% | 0 | 0 | — |
case-04 | fail→pass | 20,147 | 18,794 | -7% | 1 | 1 | 0% | 2,807 | 3,744 | +33% | 0 | 0 | — |
case-05 | fail→pass | 20,358 | 20,440 | +0% | 1 | 1 | 0% | 2,896 | 3,869 | +34% | 0 | 0 | — |
case-06 | pass→pass | 18,198 | 18,977 | +4% | 1 | 1 | 0% | 2,586 | 3,810 | +47% | 0 | 0 | — |
case-07 | pass→pass | 15,903 | 13,484 | -15% | 1 | 1 | 0% | 2,282 | 3,154 | +38% | 0 | 0 | — |
case-08 | pass→pass | 20,016 | 20,650 | +3% | 1 | 1 | 0% | 2,708 | 3,752 | +39% | 0 | 0 | — |
case-09 | pass→pass | 22,413 | 19,022 | -15% | 1 | 1 | 0% | 3,018 | 4,026 | +33% | 0 | 0 | — |
case-10 | fail→pass | 19,901 | 17,399 | -13% | 1 | 1 | 0% | 2,771 | 3,639 | +31% | 0 | 0 | — |
case-11 | fail→pass | 16,977 | 19,036 | +12% | 1 | 1 | 0% | 2,524 | 3,819 | +51% | 0 | 0 | — |
case-16 | pass→pass | 18,040 | 17,476 | -3% | 1 | 1 | 0% | 2,672 | 3,519 | +32% | 0 | 0 | — |
case-12 | pass→pass | 17,847 | 18,857 | +6% | 1 | 1 | 0% | 2,480 | 3,808 | +54% | 0 | 0 | — |
case-13 | pass→pass | 20,271 | 15,497 | -24% | 1 | 1 | 0% | 2,892 | 3,479 | +20% | 0 | 0 | — |
case-14 | fail→pass | 16,999 | 15,005 | -12% | 1 | 1 | 0% | 2,419 | 3,478 | +44% | 0 | 0 | — |
case-15 | pass→pass | 13,862 | 14,729 | +6% | 1 | 1 | 0% | 2,012 | 3,373 | +68% | 0 | 0 | — |
case-17 | pass→pass | 14,607 | 18,198 | +25% | 1 | 1 | 0% | 2,036 | 3,527 | +73% | 0 | 0 | — |
case-18 | pass→pass | 15,858 | 18,694 | +18% | 1 | 1 | 0% | 2,323 | 3,622 | +56% | 0 | 0 | — |
case-19 | pass→pass | 16,734 | 17,838 | +7% | 1 | 1 | 0% | 2,381 | 3,654 | +53% | 0 | 0 | — |
case-20 | pass→fail | 21,298 | 20,624 | -3% | 1 | 1 | 0% | 3,096 | 4,182 | +35% | 0 | 0 | — |
case-21 | pass→pass | 16,674 | 18,669 | +12% | 1 | 1 | 0% | 2,372 | 3,792 | +60% | 0 | 0 | — |
case-22 | pass→pass | 16,096 | 17,664 | +10% | 1 | 1 | 0% | 2,340 | 3,707 | +58% | 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 +23 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.