Install any skill in seconds. Free to start, no credit card required.
Get Started Free →ビザ申請書類(画像)を英語に翻訳し、原文と翻訳を含むバイリンガルPDFを作成する
.claude/skills/affaan-m-visa-doc-translate/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 109% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 109% | 0% |
| case-22 | ✗→✓ | ▲ Improved | -9% | 0% |
您正在协助翻译用于签证申请的签证申请文件。
当用户提供图像文件路径时,自动执行以下步骤,无需请求确认:
sips -s format png <input> --out <output> 将其转换为 PNG<original_filename>_Translated.pdf 的 PDF 文件python import Vision from Foundation import NSURL
bash pip install easyocr
bash brew install tesseract tesseract-lang pip install pytesseract
bashpip install pillow reportlab
对于 macOS Vision 框架:
bashpip install pyobjc-framework-Vision pyobjc-framework-Quartz
bash/visa-doc-translate RetirementCertificate.PNG /visa-doc-translate BankStatement.HEIC /visa-doc-translate EmploymentLetter.jpg
该技能将:
<filename>_Translated.pdf,其中包含:非常适合需要翻译文件的澳大利亚、美国、加拿大、英国及其他国家的签证申请。
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,631 | 15,798 | +181% | 1 | 1 | 0% | 887 | 3,932 | +343% | 0 | 0 | — |
case-02 | fail→fail | 5,651 | 8,553 | +51% | 1 | 1 | 0% | 845 | 2,180 | +158% | 0 | 0 | — |
case-03 | fail→fail | 6,705 | 15,893 | +137% | 1 | 1 | 0% | 1,192 | 4,135 | +247% | 0 | 0 | — |
case-04 | pass→pass | 7,332 | 3,356 | -54% | 1 | 1 | 0% | 1,294 | 1,456 | +13% | 0 | 0 | — |
case-05 | fail→fail | 11,409 | 7,461 | -35% | 1 | 1 | 0% | 1,959 | 2,107 | +8% | 0 | 0 | — |
case-06 | fail→pass | 17,790 | 6,462 | -64% | 1 | 1 | 0% | 2,802 | 1,907 | -32% | 0 | 0 | — |
case-07 | pass→pass | 8,069 | 5,494 | -32% | 1 | 1 | 0% | 1,289 | 1,728 | +34% | 0 | 0 | — |
case-08 | fail→fail | 10,532 | 5,117 | -51% | 1 | 1 | 0% | 1,657 | 1,708 | +3% | 0 | 0 | — |
case-09 | fail→pass | 12,590 | 3,526 | -72% | 1 | 1 | 0% | 1,968 | 1,472 | -25% | 0 | 0 | — |
case-10 | pass→pass | 23,225 | 4,766 | -79% | 1 | 1 | 0% | 2,049 | 1,725 | -16% | 0 | 0 | — |
case-11 | pass→fail | 14,159 | 5,159 | -64% | 1 | 1 | 0% | 2,512 | 1,600 | -36% | 0 | 0 | — |
case-12 | fail→pass | 4,143 | 2,536 | -39% | 1 | 1 | 0% | 632 | 1,324 | +109% | 0 | 0 | — |
case-13 | fail→fail | 9,208 | 3,041 | -67% | 1 | 1 | 0% | 1,306 | 1,387 | +6% | 0 | 0 | — |
case-14 | fail→fail | 9,184 | 6,506 | -29% | 1 | 1 | 0% | 1,497 | 1,928 | +29% | 0 | 0 | — |
case-15 | pass→pass | 9,506 | 4,797 | -50% | 1 | 1 | 0% | 1,544 | 1,603 | +4% | 0 | 0 | — |
case-16 | pass→pass | 4,980 | 1,900 | -62% | 1 | 1 | 0% | 711 | 1,219 | +71% | 0 | 0 | — |
case-17 | fail→fail | 13,345 | 8,352 | -37% | 1 | 1 | 0% | 2,315 | 2,444 | +6% | 0 | 0 | — |
case-18 | fail→pass | 4,795 | 4,168 | -13% | 1 | 1 | 0% | 691 | 1,446 | +109% | 0 | 0 | — |
case-19 | pass→pass | 8,257 | 4,649 | -44% | 1 | 1 | 0% | 1,272 | 1,495 | +18% | 0 | 0 | — |
case-20 | pass→pass | 5,172 | 3,841 | -26% | 1 | 1 | 0% | 882 | 1,503 | +70% | 0 | 0 | — |
case-21 | fail→fail | 11,147 | 5,548 | -50% | 1 | 1 | 0% | 1,752 | 1,713 | -2% | 0 | 0 | — |
case-22 | fail→pass | 10,724 | 5,166 | -52% | 1 | 1 | 0% | 1,903 | 1,736 | -9% | 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 +18 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.