Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →定义小芽家教 AI 助手的核心人格特质、教学风格和语言规范，确保"温柔耐心、会引导、不给答案"的教学理念。
.claude/skills/majiayu000-sprout-persona/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 128% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 34% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -12% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 19% | 0% |
| 场景 | 应该用 | 不应该用 | |------|--------|----------| | 学生答错 | "没关系，我们再来想一想！" | "错了，再想想" | | 学生答对 | "太棒了！你找到方法了！" | "对了" | | 学生求助 | "别着急，小芽老师陪你一起看" | "这个很简单的" |
pythonPROHIBITED_WORDS = ["简单", "容易", "笨", "傻", "怎么还不会", "不对", "错了"] ENCOURAGEMENT_WORDS = ["很棒", "有进步", "想得很好", "加油", "没关系", "再试试"]
绝对禁止：
必须做到：
tdd-cycle - TDD 开发流程teaching-strategy - 问题类型识别socratic-teaching - 引导式教学| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 13,475 | 19,140 | +42% | 1 | 1 | 0% | 1,858 | 2,416 | +30% | 0 | 0 | — |
case-07 | fail→pass | 5,504 | 20,228 | +268% | 1 | 1 | 0% | 854 | 1,943 | +128% | 0 | 0 | — |
case-13 | fail→fail | 11,771 | 13,621 | +16% | 1 | 1 | 0% | 1,033 | 1,739 | +68% | 0 | 0 | — |
case-02 | fail→fail | 8,735 | 14,206 | +63% | 1 | 1 | 0% | 1,295 | 1,954 | +51% | 0 | 0 | — |
case-03 | fail→pass | 24,019 | 10,771 | -55% | 1 | 1 | 0% | 1,694 | 1,911 | +13% | 0 | 0 | — |
case-04 | fail→pass | 11,672 | 16,162 | +38% | 1 | 1 | 0% | 1,504 | 2,014 | +34% | 0 | 0 | — |
case-05 | fail→pass | 19,012 | 13,371 | -30% | 1 | 1 | 0% | 1,965 | 1,732 | -12% | 0 | 0 | — |
case-06 | fail→pass | 10,580 | 8,654 | -18% | 1 | 1 | 0% | 1,375 | 1,633 | +19% | 0 | 0 | — |
case-08 | fail→fail | 16,055 | 18,939 | +18% | 1 | 1 | 0% | 1,743 | 1,932 | +11% | 0 | 0 | — |
case-09 | fail→pass | 15,404 | 16,801 | +9% | 1 | 1 | 0% | 1,982 | 2,147 | +8% | 0 | 0 | — |
case-10 | fail→pass | 15,732 | 14,693 | -7% | 1 | 1 | 0% | 1,586 | 1,918 | +21% | 0 | 0 | — |
case-11 | fail→pass | 13,304 | 13,861 | +4% | 1 | 1 | 0% | 1,310 | 1,753 | +34% | 0 | 0 | — |
case-12 | fail→pass | 10,300 | 8,126 | -21% | 1 | 1 | 0% | 1,430 | 1,601 | +12% | 0 | 0 | — |
case-14 | fail→fail | 14,004 | 13,180 | -6% | 1 | 1 | 0% | 1,204 | 1,674 | +39% | 0 | 0 | — |
case-15 | fail→pass | 13,167 | 17,390 | +32% | 1 | 1 | 0% | 1,285 | 1,881 | +46% | 0 | 0 | — |
case-16 | fail→fail | 14,964 | 32,304 | +116% | 1 | 1 | 0% | 1,706 | 1,585 | -7% | 0 | 0 | — |
case-17 | fail→fail | 12,437 | 13,932 | +12% | 1 | 1 | 0% | 1,052 | 1,698 | +61% | 0 | 0 | — |
case-18 | pass→pass | 12,660 | 13,049 | +3% | 1 | 1 | 0% | 1,654 | 1,575 | -5% | 0 | 0 | — |
case-19 | pass→fail | 7,746 | 10,829 | +40% | 1 | 1 | 0% | 417 | 1,929 | +363% | 0 | 0 | — |
case-20 | pass→pass | 22,639 | 22,189 | -2% | 1 | 1 | 0% | 3,008 | 3,097 | +3% | 0 | 0 | — |
case-21 | pass→pass | 18,126 | 19,732 | +9% | 1 | 1 | 0% | 2,659 | 2,671 | +0% | 0 | 0 | — |
case-22 | pass→fail | 14,494 | 22,215 | +53% | 1 | 1 | 0% | 2,283 | 2,867 | +26% | 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 +36 percentage points is the difference between those two pass rates over the 22 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.