Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Umbrella skill for document workflows (PDF/DOCX/XLSX/PPTX). Dispatches to the most specific document skill to reduce noise and improve routing precision.
.claude/skills/foryourhealth111-pixel-document-skills/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 36% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -25% | 0% |
Use this skill when the task is clearly “document work” but the exact format is not yet fixed, or when the user mixes multiple formats (e.g., “把论文里的表格做成 Excel，再导出 PDF 报告”).
Goal: fast dispatch to the most specific skill so we keep high hit-rate / low noise.
1) PDF (.pdf, “PDF”, “pypdf”, “pdfplumber”, “render pages”)
2) Word / DOCX (.docx, “Word”, “tracked changes”, “python-docx”)
.docx formatting/layout heavy and the doc skill is requested/required by your environment).3) Excel / Spreadsheets (.xlsx, .csv, .tsv, “Excel”, “openpyxl”, “pivot table”)
4) Slides / Posters / PPTX (.pptx, “slides”, “poster”, “deck”, “PowerPoint”)
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 7,870 | 3,002 | -62% | 1 | 1 | 0% | 1,402 | 947 | -32% | 0 | 0 | — |
case-02 | fail→pass | 7,098 | 3,252 | -54% | 1 | 1 | 0% | 1,149 | 1,054 | -8% | 0 | 0 | — |
case-03 | fail→pass | 5,069 | 3,990 | -21% | 1 | 1 | 0% | 866 | 1,180 | +36% | 0 | 0 | — |
case-04 | fail→pass | 5,489 | 2,148 | -61% | 1 | 1 | 0% | 833 | 914 | +10% | 0 | 0 | — |
case-05 | fail→pass | 10,671 | 3,628 | -66% | 1 | 1 | 0% | 1,515 | 1,143 | -25% | 0 | 0 | — |
case-06 | fail→pass | 8,750 | 3,162 | -64% | 1 | 1 | 0% | 1,494 | 1,064 | -29% | 0 | 0 | — |
case-07 | pass→pass | 4,141 | 2,816 | -32% | 1 | 1 | 0% | 610 | 971 | +59% | 0 | 0 | — |
case-08 | fail→pass | 6,128 | 2,702 | -56% | 1 | 1 | 0% | 997 | 908 | -9% | 0 | 0 | — |
case-09 | pass→pass | 8,831 | 7,620 | -14% | 1 | 1 | 0% | 1,568 | 1,805 | +15% | 0 | 0 | — |
case-10 | fail→pass | 9,821 | 5,097 | -48% | 1 | 1 | 0% | 1,695 | 1,375 | -19% | 0 | 0 | — |
case-11 | pass→pass | 12,319 | 8,819 | -28% | 1 | 1 | 0% | 1,904 | 1,851 | -3% | 0 | 0 | — |
case-12 | fail→pass | 6,028 | 2,359 | -61% | 1 | 1 | 0% | 1,134 | 932 | -18% | 0 | 0 | — |
case-13 | pass→pass | 13,540 | 2,250 | -83% | 1 | 1 | 0% | 1,394 | 843 | -40% | 0 | 0 | — |
case-14 | fail→pass | 6,593 | 3,623 | -45% | 1 | 1 | 0% | 1,047 | 1,071 | +2% | 0 | 0 | — |
case-15 | fail→pass | 7,429 | 4,378 | -41% | 1 | 1 | 0% | 1,216 | 1,199 | -1% | 0 | 0 | — |
case-16 | pass→pass | 5,083 | 3,687 | -27% | 1 | 1 | 0% | 828 | 1,027 | +24% | 0 | 0 | — |
case-17 | pass→pass | 12,577 | 6,961 | -45% | 1 | 1 | 0% | 2,185 | 1,644 | -25% | 0 | 0 | — |
case-18 | pass→pass | 9,328 | 6,046 | -35% | 1 | 1 | 0% | 1,251 | 1,466 | +17% | 0 | 0 | — |
case-19 | pass→pass | 5,962 | 4,352 | -27% | 1 | 1 | 0% | 959 | 1,262 | +32% | 0 | 0 | — |
case-20 | pass→pass | 5,939 | 6,348 | +7% | 1 | 1 | 0% | 1,238 | 1,785 | +44% | 0 | 0 | — |
case-21 | pass→pass | 5,400 | 2,951 | -45% | 1 | 1 | 0% | 903 | 929 | +3% | 0 | 0 | — |
case-22 | pass→pass | 5,741 | 4,197 | -27% | 1 | 1 | 0% | 1,072 | 1,298 | +21% | 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 +50 percentage points is the difference between those two pass rates over the 22 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.