Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →大きなPDFファイルを圧縮するスキル。ファイルサイズを最大98%削減。 「PDFを圧縮して」「PDFを軽くして」「ファイルサイズを小さくして」等のリクエストで発動。
.claude/skills/minicoohei-pdf-compressor/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-16 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -4% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 212% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -33% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -56% | 0% |
Compress PDF files by converting pages to optimized images and rebuilding the PDF.
bashpython scripts/compress.py "{pdf_path}" --width {width} --quality {quality} --output "{output_path}"
| Parameter | Required | Default | Description | |-----------|----------|---------|-------------| | pdf_path | Yes | - | Path to PDF file to compress | | --width | No | 1920 | Page width in pixels | | --quality | No | 85 | JPEG quality (1-100) | | --dpi | No | 150 | DPI for PDF to image conversion | | --output, -o | No | auto | Output path (default: {filename}_compressed.pdf) |
| Use Case | Width | Quality | Expected Reduction | |----------|-------|---------|-------------------| | Web/Email | 1280 | 75 | ~95% | | Standard | 1920 | 85 | ~90% | | High Quality | 2560 | 90 | ~80% | | Print | 3840 | 95 | ~60% |
bash# Basic compression (default settings) python scripts/compress.py "large_presentation.pdf" # Web-optimized (smaller file) python scripts/compress.py "slides.pdf" --width 1280 --quality 75 # High quality for presentations python scripts/compress.py "report.pdf" --width 2560 --quality 90 # Custom output path python scripts/compress.py "document.pdf" -o "document_small.pdf"
brew install popplerapt-get install poppler-utils| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-16 | fail→pass | 11,197 | 2,042 | -82% | 1 | 1 | 0% | 1,856 | 928 | -50% | 0 | 0 | — |
case-01 | fail→pass | 11,337 | 11,910 | +5% | 1 | 1 | 0% | 2,100 | 2,008 | -4% | 0 | 0 | — |
case-10 | pass→pass | 2,606 | 2,558 | -2% | 1 | 1 | 0% | 429 | 911 | +112% | 0 | 0 | — |
case-02 | fail→pass | 8,226 | 5,363 | -35% | 1 | 1 | 0% | 297 | 926 | +212% | 0 | 0 | — |
case-03 | pass→pass | 13,757 | 7,795 | -43% | 1 | 1 | 0% | 2,271 | 2,014 | -11% | 0 | 0 | — |
case-04 | pass→fail | 8,406 | 5,551 | -34% | 1 | 1 | 0% | 1,587 | 1,721 | +8% | 0 | 0 | — |
case-05 | pass→pass | 9,754 | 6,306 | -35% | 1 | 1 | 0% | 1,884 | 1,826 | -3% | 0 | 0 | — |
case-06 | fail→pass | 8,460 | 2,233 | -74% | 1 | 1 | 0% | 1,487 | 997 | -33% | 0 | 0 | — |
case-07 | fail→pass | 11,646 | 2,120 | -82% | 1 | 1 | 0% | 2,002 | 890 | -56% | 0 | 0 | — |
case-08 | pass→pass | 6,526 | 2,434 | -63% | 1 | 1 | 0% | 1,185 | 966 | -18% | 0 | 0 | — |
case-09 | fail→pass | 3,602 | 2,362 | -34% | 1 | 1 | 0% | 552 | 925 | +68% | 0 | 0 | — |
case-11 | pass→pass | 2,382 | 1,681 | -29% | 1 | 1 | 0% | 371 | 794 | +114% | 0 | 0 | — |
case-12 | pass→pass | 3,288 | 3,048 | -7% | 1 | 1 | 0% | 519 | 1,015 | +96% | 0 | 0 | — |
case-13 | pass→pass | 8,008 | 2,239 | -72% | 1 | 1 | 0% | 1,400 | 902 | -36% | 0 | 0 | — |
case-14 | fail→pass | 7,485 | 2,030 | -73% | 1 | 1 | 0% | 1,170 | 898 | -23% | 0 | 0 | — |
case-15 | pass→pass | 9,688 | 3,700 | -62% | 1 | 1 | 0% | 1,482 | 1,126 | -24% | 0 | 0 | — |
case-17 | fail→pass | 4,876 | 2,770 | -43% | 1 | 1 | 0% | 898 | 989 | +10% | 0 | 0 | — |
case-18 | fail→pass | 7,267 | 1,825 | -75% | 1 | 1 | 0% | 1,144 | 824 | -28% | 0 | 0 | — |
case-19 | fail→pass | 9,923 | 1,536 | -85% | 1 | 1 | 0% | 1,568 | 802 | -49% | 0 | 0 | — |
case-20 | fail→pass | 9,485 | 1,731 | -82% | 1 | 1 | 0% | 1,553 | 867 | -44% | 0 | 0 | — |
case-21 | pass→pass | 11,331 | 2,831 | -75% | 1 | 1 | 0% | 1,646 | 1,007 | -39% | 0 | 0 | — |
case-22 | fail→pass | 5,186 | 1,513 | -71% | 1 | 1 | 0% | 930 | 756 | -19% | 0 | 0 | — |
case-23 | fail→pass | 10,391 | 1,976 | -81% | 1 | 1 | 0% | 1,856 | 902 | -51% | 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. 23 cases were attempted, and 22 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 +52 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.