Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →スクリーンショットからエラー診断や操作チュートリアルを自動生成するスキル。 「スクショを分析して」「画面のエラーを調べて」「操作手順を作って」等のリクエストで発動。
.claude/skills/minicoohei-screenshot-analyzer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | -72% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 167% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -67% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -43% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -49% | 0% |
Analyze screenshots for error diagnosis or operation tutorial generation.
bash# Error analysis (default) python scripts/analyze.py "{screenshot}" --mode analyze # Tutorial generation python scripts/analyze.py "{screenshot}" --mode tutorial
| Parameter | Required | Default | Description | |-----------|----------|---------|-------------| | screenshot | Yes | - | Path to screenshot | | --mode, -m | No | analyze | Mode: analyze or tutorial | | --output, -o | No | auto | Output HTML path | | --no-annotate | No | false | Skip annotation generation |
bash# Analyze error screenshot python scripts/analyze.py "error_console.png" # Generate tutorial python scripts/analyze.py "settings_menu.png" --mode tutorial # Specify output python scripts/analyze.py "login.png" --mode tutorial --output "docs/login_guide.html" # Skip annotations python scripts/analyze.py "error.png" --no-annotate
スクリーンショットからエラー診断や操作チュートリアルを自動生成するスキルです。Gemini Vision APIでUI要素を認識し、エラーの原因分析・解決策提案、またはステップバイステップの操作ガイドを出力します。
| エラー | 解決方法 | |--------|---------| | API key not found | GEMINI_API_KEY または GOOGLE_API_KEY を環境変数に設定 | | Analysis returned empty results | スクリーンショットが不鮮明な可能性。より高解像度の画像を使用 |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-13 | fail→pass | 13,381 | 1,725 | -87% | 1 | 1 | 0% | 2,931 | 832 | -72% | 0 | 0 | — |
case-01 | fail→fail | 13,764 | 2,874 | -79% | 1 | 1 | 0% | 337 | 853 | +153% | 0 | 0 | — |
case-02 | fail→pass | 3,701 | 7,246 | +96% | 1 | 1 | 0% | 280 | 748 | +167% | 0 | 0 | — |
case-03 | fail→pass | 14,383 | 3,318 | -77% | 1 | 1 | 0% | 2,488 | 810 | -67% | 0 | 0 | — |
case-04 | fail→pass | 9,256 | 3,587 | -61% | 1 | 1 | 0% | 1,556 | 888 | -43% | 0 | 0 | — |
case-05 | fail→pass | 10,234 | 2,575 | -75% | 1 | 1 | 0% | 1,897 | 970 | -49% | 0 | 0 | — |
case-06 | pass→pass | 13,263 | 2,411 | -82% | 1 | 1 | 0% | 2,147 | 893 | -58% | 0 | 0 | — |
case-07 | pass→pass | 3,549 | 1,836 | -48% | 1 | 1 | 0% | 531 | 825 | +55% | 0 | 0 | — |
case-08 | fail→pass | 5,616 | 1,927 | -66% | 1 | 1 | 0% | 915 | 787 | -14% | 0 | 0 | — |
case-09 | pass→pass | 7,810 | 1,391 | -82% | 1 | 1 | 0% | 631 | 742 | +18% | 0 | 0 | — |
case-10 | pass→pass | 4,048 | 1,776 | -56% | 1 | 1 | 0% | 605 | 768 | +27% | 0 | 0 | — |
case-11 | pass→pass | 13,160 | 2,112 | -84% | 1 | 1 | 0% | 2,023 | 883 | -56% | 0 | 0 | — |
case-12 | pass→pass | 9,181 | 2,358 | -74% | 1 | 1 | 0% | 1,378 | 921 | -33% | 0 | 0 | — |
case-14 | fail→pass | 15,307 | 1,802 | -88% | 1 | 1 | 0% | 2,369 | 822 | -65% | 0 | 0 | — |
case-15 | fail→pass | 8,854 | 3,930 | -56% | 1 | 1 | 0% | 1,595 | 1,234 | -23% | 0 | 0 | — |
case-16 | fail→pass | 11,059 | 1,908 | -83% | 1 | 1 | 0% | 1,798 | 842 | -53% | 0 | 0 | — |
case-17 | fail→pass | 6,335 | 4,269 | -33% | 1 | 1 | 0% | 1,143 | 1,258 | +10% | 0 | 0 | — |
case-18 | fail→pass | 11,385 | 1,597 | -86% | 1 | 1 | 0% | 1,824 | 769 | -58% | 0 | 0 | — |
case-19 | pass→pass | 5,660 | 1,653 | -71% | 1 | 1 | 0% | 853 | 758 | -11% | 0 | 0 | — |
case-20 | pass→pass | 13,644 | 13,198 | -3% | 1 | 1 | 0% | 2,693 | 3,453 | +28% | 0 | 0 | — |
case-21 | pass→pass | 17,425 | 16,942 | -3% | 1 | 1 | 0% | 3,851 | 3,994 | +4% | 0 | 0 | — |
case-22 | pass→pass | 13,012 | 12,094 | -7% | 1 | 1 | 0% | 2,523 | 2,723 | +8% | 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 +50 percentage points is the difference between those two pass rates over the 21 comparable cases.
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.