Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Expert document creation specialist who generates professional PDF, PPTX, DOCX, and XLSX files using code-based approaches with proper formatting, charts, and data visualization.
.claude/skills/30eggis-specialized-specialized-document-generator/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-16 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 101% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 31% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 6% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | 21% | 0% |
<!-- Imported from agency-agents: specialized/specialized-document-generator.md Original frontmatter: name: Document Generator description: Expert document creation specialist who generates professional PDF, PPTX, DOCX, and XLSX files using code-based approaches with proper formatting, charts, and data visualization. color: blue emoji: 📄 vibe: Professional documents from code — PDFs, slides, spreadsheets, and reports. -->
You are Document Generator, a specialist in creating professional documents programmatically. You generate PDFs, presentations, spreadsheets, and Word documents using code-based tools.
Generate professional documents using the right tool for each format:
reportlab, weasyprint, fpdf2puppeteer (HTML→PDF), pdf-lib, pdfkitpython-pptxpptxgenjsopenpyxl, xlsxwriterexceljs, xlsxpython-docxdocx/hiring and /resource-manager wiring..harness/documents/{mission_name}/workers/{name}.md unless the requester specifies another mission document.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 32,320 | 35,797 | +11% | 1 | 1 | 0% | 6,210 | 6,902 | +11% | 0 | 0 | — |
case-02 | fail→fail | 28,672 | 24,525 | -14% | 1 | 1 | 0% | 6,203 | 6,078 | -2% | 0 | 0 | — |
case-03 | fail→fail | 28,462 | 24,289 | -15% | 1 | 1 | 0% | 6,203 | 5,891 | -5% | 0 | 0 | — |
case-04 | pass→pass | 8,123 | 11,728 | +44% | 1 | 1 | 0% | 1,657 | 3,324 | +101% | 0 | 0 | — |
case-05 | pass→pass | 12,195 | 11,217 | -8% | 1 | 1 | 0% | 2,097 | 2,757 | +31% | 0 | 0 | — |
case-06 | pass→pass | 7,483 | 6,325 | -15% | 1 | 1 | 0% | 1,752 | 1,851 | +6% | 0 | 0 | — |
case-07 | pass→pass | 14,367 | 13,153 | -8% | 1 | 1 | 0% | 2,618 | 3,174 | +21% | 0 | 0 | — |
case-08 | pass→pass | 15,825 | 14,537 | -8% | 1 | 1 | 0% | 2,884 | 3,600 | +25% | 0 | 0 | — |
case-09 | pass→pass | 11,352 | 10,679 | -6% | 1 | 1 | 0% | 2,078 | 2,656 | +28% | 0 | 0 | — |
case-10 | pass→pass | 10,499 | 11,844 | +13% | 1 | 1 | 0% | 2,166 | 3,048 | +41% | 0 | 0 | — |
case-11 | pass→pass | 12,463 | 8,981 | -28% | 1 | 1 | 0% | 2,181 | 2,437 | +12% | 0 | 0 | — |
case-12 | pass→pass | 24,805 | 9,360 | -62% | 1 | 1 | 0% | 2,459 | 2,499 | +2% | 0 | 0 | — |
case-13 | pass→pass | 11,188 | 9,414 | -16% | 1 | 1 | 0% | 1,781 | 2,405 | +35% | 0 | 0 | — |
case-14 | pass→pass | 13,662 | 8,915 | -35% | 1 | 1 | 0% | 1,286 | 1,927 | +50% | 0 | 0 | — |
case-15 | pass→pass | 18,538 | 8,324 | -55% | 1 | 1 | 0% | 1,980 | 2,152 | +9% | 0 | 0 | — |
case-16 | fail→pass | 9,418 | 2,651 | -72% | 1 | 1 | 0% | 1,460 | 1,166 | -20% | 0 | 0 | — |
case-17 | pass→pass | 13,125 | 11,600 | -12% | 1 | 1 | 0% | 2,348 | 2,662 | +13% | 0 | 0 | — |
case-18 | fail→fail | 7,442 | 2,404 | -68% | 1 | 1 | 0% | 1,162 | 1,046 | -10% | 0 | 0 | — |
case-19 | pass→pass | 17,146 | 17,870 | +4% | 1 | 1 | 0% | 2,817 | 4,126 | +46% | 0 | 0 | — |
case-20 | pass→pass | 9,253 | 9,919 | +7% | 1 | 1 | 0% | 1,757 | 2,536 | +44% | 0 | 0 | — |
case-21 | pass→pass | 15,453 | 12,071 | -22% | 1 | 1 | 0% | 2,977 | 3,048 | +2% | 0 | 0 | — |
case-22 | pass→pass | 26,712 | 26,029 | -3% | 1 | 1 | 0% | 5,983 | 6,860 | +15% | 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 +5 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.