Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →自动化生成历史科学类3分钟科普短视频的全流程素材包，包含口播文案、分镜脚本、Veo2提示词、人物形象规范等，适配即梦平台视频生成。
.claude/skills/anbeime-historical-science-video-prod/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-12 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 61% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -28% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 42% | 0% |
scripts/generate_video_materials.py 处理...output/script.txt - 3分钟口播文案（约900-1100字）output/storyboard.md - 分镜脚本表（30-35个分镜）output/veo2_prompts.txt - Veo2提示词清单output/character_design.md - 人物形象规范bashpython scripts/generate_video_materials.py \ --theme "蝴蝶效应的起源" \ --era "1961年" \ --core_conclusion "初始条件的微小变化会导致巨大差异，无法长期预测天气" \ --output_dir "./output"
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-12 | fail→pass | 20,288 | 1,814 | -91% | 1 | 1 | 0% | 892 | 1,076 | +21% | 0 | 0 | — |
case-01 | fail→fail | 23,903 | 35,329 | +48% | 1 | 1 | 0% | 3,986 | 6,554 | +64% | 0 | 0 | — |
case-02 | fail→fail | 22,638 | 32,733 | +45% | 1 | 1 | 0% | 3,617 | 6,123 | +69% | 0 | 0 | — |
case-18 | fail→pass | 7,115 | 5,723 | -20% | 1 | 1 | 0% | 1,135 | 1,693 | +49% | 0 | 0 | — |
case-03 | fail→fail | 29,696 | 35,304 | +19% | 1 | 1 | 0% | 3,570 | 6,792 | +90% | 0 | 0 | — |
case-04 | fail→pass | 6,975 | 5,344 | -23% | 1 | 1 | 0% | 1,177 | 1,897 | +61% | 0 | 0 | — |
case-05 | fail→pass | 12,482 | 5,525 | -56% | 1 | 1 | 0% | 2,105 | 1,511 | -28% | 0 | 0 | — |
case-06 | fail→fail | 18,825 | 14,468 | -23% | 1 | 1 | 0% | 3,351 | 3,494 | +4% | 0 | 0 | — |
case-07 | fail→fail | 13,239 | 6,966 | -47% | 1 | 1 | 0% | 1,967 | 1,930 | -2% | 0 | 0 | — |
case-08 | fail→pass | 11,040 | 10,524 | -5% | 1 | 1 | 0% | 2,011 | 2,847 | +42% | 0 | 0 | — |
case-09 | fail→pass | 12,558 | 2,251 | -82% | 1 | 1 | 0% | 2,085 | 1,224 | -41% | 0 | 0 | — |
case-10 | fail→pass | 9,202 | 4,678 | -49% | 1 | 1 | 0% | 1,648 | 1,478 | -10% | 0 | 0 | — |
case-11 | fail→pass | 14,441 | 2,290 | -84% | 1 | 1 | 0% | 2,369 | 1,246 | -47% | 0 | 0 | — |
case-13 | fail→fail | 7,600 | 1,635 | -78% | 1 | 1 | 0% | 1,470 | 1,163 | -21% | 0 | 0 | — |
case-14 | pass→pass | 10,301 | 2,249 | -78% | 1 | 1 | 0% | 1,806 | 1,197 | -34% | 0 | 0 | — |
case-15 | pass→pass | 12,325 | 14,257 | +16% | 1 | 1 | 0% | 2,166 | 2,771 | +28% | 0 | 0 | — |
case-16 | fail→pass | 11,696 | 2,972 | -75% | 1 | 1 | 0% | 1,562 | 1,358 | -13% | 0 | 0 | — |
case-17 | fail→pass | 7,455 | 2,672 | -64% | 1 | 1 | 0% | 1,228 | 1,268 | +3% | 0 | 0 | — |
case-19 | fail→fail | 8,853 | 2,481 | -72% | 1 | 1 | 0% | 1,587 | 1,241 | -22% | 0 | 0 | — |
case-20 | fail→fail | 12,507 | 16,528 | +32% | 1 | 1 | 0% | 2,593 | 4,511 | +74% | 0 | 0 | — |
case-21 | pass→fail | 8,559 | 18,572 | +117% | 1 | 1 | 0% | 1,688 | 4,156 | +146% | 0 | 0 | — |
case-22 | fail→pass | 22,342 | 8,999 | -60% | 1 | 1 | 0% | 4,499 | 2,377 | -47% | 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 +45 percentage points is the difference between those two pass rates over the 21 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.