Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →为3-12岁儿童提供温馨亲切的睡前寓言故事和成语典故讲解。支持用户唤醒后提供故事列表选择，或直接讲解指定故事/成语。讲解时保持亲切温馨的语气、0.6倍正常语速、通俗易懂的表达，为小朋友营造舒适的睡前氛围。
.claude/skills/anbeime-bedtime-story/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 88% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 770% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 542% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 188% | 0% |
触发条件：
处理逻辑：
当用户发送"我要听故事"时：
推送规则（强制执行）：
列表结构参考：
1. 《狐狸和葡萄》——《伊索寓言》
2. 《龟兔赛跑》——《伊索寓言》
...等待用户选择：
讲解要求（强制）：
讲解要求（强制）：
用户：我要听故事
智能体：小朋友，这里有20个好听的寓言故事，你想听第几个呀？
1. 《狐狸和葡萄》——《伊索寓言》
2. 《龟兔赛跑》——《伊索寓言》
...
20. 《守株待兔》——《韩非子》（寓言版）用户：3
智能体：小朋友们好呀，今天我们一起来听《乌鸦喝水》的故事～
一只乌鸦口渴了，到处找水喝...
（全文讲解，语速0.6倍，亲切温馨）
好啦，故事讲完了，晚安宝贝们～用户：3、7
智能体：好的，我们一起来听这两个故事～
第一个故事：《乌鸦喝水》
...
第二个故事：《东郭先生和狼》
...
（两个故事连播）用户：守株待兔
智能体：小朋友们好呀，今天我们一起来学习成语"守株待兔"～
这个成语有一个有趣的典故...
（讲解成语典故、含义、用法）
现在你们知道"守株待兔"是什么意思了吗？用户：不喜欢
智能体：没关系，那我们换一批故事～
（生成新的20个故事列表）
小朋友，这里有20个好听的寓言故事，你想听第几个呀？| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | fail→pass | 10,877 | 12,027 | +11% | 1 | 1 | 0% | 1,780 | 3,340 | +88% | 0 | 0 | — |
case-03 | fail→pass | 3,416 | 13,487 | +295% | 1 | 1 | 0% | 464 | 4,039 | +770% | 0 | 0 | — |
case-09 | fail→fail | 15,967 | 6,168 | -61% | 1 | 1 | 0% | 2,541 | 2,754 | +8% | 0 | 0 | — |
case-01 | fail→pass | 14,497 | 6,696 | -54% | 1 | 1 | 0% | 2,575 | 2,784 | +8% | 0 | 0 | — |
case-04 | fail→pass | 5,215 | 18,079 | +247% | 1 | 1 | 0% | 781 | 5,015 | +542% | 0 | 0 | — |
case-05 | fail→pass | 5,815 | 5,943 | +2% | 1 | 1 | 0% | 929 | 2,671 | +188% | 0 | 0 | — |
case-06 | pass→pass | 13,423 | 10,154 | -24% | 1 | 1 | 0% | 2,029 | 3,396 | +67% | 0 | 0 | — |
case-07 | fail→pass | 9,535 | 12,448 | +31% | 1 | 1 | 0% | 1,140 | 3,412 | +199% | 0 | 0 | — |
case-08 | fail→pass | 3,151 | 11,223 | +256% | 1 | 1 | 0% | 426 | 3,122 | +633% | 0 | 0 | — |
case-10 | fail→fail | 6,375 | 7,964 | +25% | 1 | 1 | 0% | 787 | 2,427 | +208% | 0 | 0 | — |
case-11 | pass→pass | 15,092 | 8,699 | -42% | 1 | 1 | 0% | 1,973 | 2,855 | +45% | 0 | 0 | — |
case-12 | pass→pass | 15,323 | 12,936 | -16% | 1 | 1 | 0% | 2,134 | 3,031 | +42% | 0 | 0 | — |
case-13 | pass→pass | 17,182 | 7,333 | -57% | 1 | 1 | 0% | 3,016 | 3,028 | +0% | 0 | 0 | — |
case-14 | fail→pass | 9,229 | 5,639 | -39% | 1 | 1 | 0% | 1,425 | 2,602 | +83% | 0 | 0 | — |
case-15 | fail→pass | 4,359 | 11,404 | +162% | 1 | 1 | 0% | 626 | 3,661 | +485% | 0 | 0 | — |
case-16 | pass→pass | 12,979 | 11,537 | -11% | 1 | 1 | 0% | 1,814 | 2,979 | +64% | 0 | 0 | — |
case-17 | pass→pass | 8,977 | 9,115 | +2% | 1 | 1 | 0% | 1,305 | 2,900 | +122% | 0 | 0 | — |
case-18 | fail→pass | 16,655 | 16,495 | -1% | 1 | 1 | 0% | 2,420 | 4,053 | +67% | 0 | 0 | — |
case-19 | fail→pass | 12,607 | 6,007 | -52% | 1 | 1 | 0% | 2,118 | 2,689 | +27% | 0 | 0 | — |
case-20 | pass→fail | 24,076 | 22,152 | -8% | 1 | 1 | 0% | 4,116 | 4,452 | +8% | 0 | 0 | — |
case-21 | pass→pass | 15,647 | 11,900 | -24% | 1 | 1 | 0% | 2,850 | 3,640 | +28% | 0 | 0 | — |
case-22 | pass→pass | 8,819 | 8,857 | +0% | 1 | 1 | 0% | 1,738 | 3,121 | +80% | 0 | 0 | — |
case-23 | fail→pass | 16,163 | 7,493 | -54% | 1 | 1 | 0% | 2,617 | 2,781 | +6% | 0 | 0 | — |
case-24 | pass→pass | 13,452 | 13,474 | +0% | 1 | 1 | 0% | 2,135 | 2,913 | +36% | 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. The headline lift of +46 percentage points is the difference between those two pass rates over the 24 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.