Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →A spreadsheet helper that converts CSV files into formatted XLSX workbooks with styled headers and frozen panes.
.claude/skills/sayed3li97-csv-to-xlsx/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 25% | 0% |
CSV is a widely used data format. This skill turns CSV input into styled XLSX workbooks.
For the delimiter you can use a comma, or a semicolon, or a tab, or any single character the source uses.
csvname,amount,date Widgets,120,2031-01-15
The output keeps one sheet per input file and uses the file name as the sheet name.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 28,225 | 3,007 | -89% | 1 | 1 | 0% | 6,176 | 553 | -91% | 0 | 0 | — |
case-02 | fail→pass | 23,050 | 9,350 | -59% | 1 | 1 | 0% | 4,929 | 2,207 | -55% | 0 | 0 | — |
case-03 | fail→fail | 26,796 | 2,800 | -90% | 1 | 1 | 0% | 6,165 | 553 | -91% | 0 | 0 | — |
case-04 | pass→pass | 8,568 | 6,620 | -23% | 1 | 1 | 0% | 1,655 | 1,443 | -13% | 0 | 0 | — |
case-05 | pass→pass | 11,985 | 10,334 | -14% | 1 | 1 | 0% | 2,437 | 2,175 | -11% | 0 | 0 | — |
case-06 | pass→pass | 15,020 | 17,016 | +13% | 1 | 1 | 0% | 3,083 | 3,453 | +12% | 0 | 0 | — |
case-07 | fail→pass | 5,308 | 7,079 | +33% | 1 | 1 | 0% | 1,026 | 1,683 | +64% | 0 | 0 | — |
case-08 | fail→pass | 8,021 | 7,548 | -6% | 1 | 1 | 0% | 1,655 | 1,762 | +6% | 0 | 0 | — |
case-09 | pass→pass | 12,282 | 6,598 | -46% | 1 | 1 | 0% | 1,938 | 1,559 | -20% | 0 | 0 | — |
case-10 | fail→pass | 5,633 | 5,728 | +2% | 1 | 1 | 0% | 1,116 | 1,291 | +16% | 0 | 0 | — |
case-11 | pass→pass | 8,247 | 3,722 | -55% | 1 | 1 | 0% | 1,710 | 856 | -50% | 0 | 0 | — |
case-12 | fail→fail | 9,290 | 8,735 | -6% | 1 | 1 | 0% | 1,789 | 812 | -55% | 0 | 0 | — |
case-13 | fail→pass | 7,672 | 7,844 | +2% | 1 | 1 | 0% | 1,568 | 1,956 | +25% | 0 | 0 | — |
case-14 | fail→pass | 9,643 | 7,526 | -22% | 1 | 1 | 0% | 1,847 | 1,849 | +0% | 0 | 0 | — |
case-15 | fail→pass | 7,662 | 6,406 | -16% | 1 | 1 | 0% | 1,467 | 1,536 | +5% | 0 | 0 | — |
case-16 | pass→pass | 8,613 | 11,863 | +38% | 1 | 1 | 0% | 1,897 | 2,955 | +56% | 0 | 0 | — |
case-17 | fail→pass | 9,189 | 6,618 | -28% | 1 | 1 | 0% | 1,955 | 1,541 | -21% | 0 | 0 | — |
case-18 | fail→pass | 9,141 | 8,380 | -8% | 1 | 1 | 0% | 1,801 | 2,044 | +13% | 0 | 0 | — |
case-19 | fail→fail | 11,206 | 2,557 | -77% | 1 | 1 | 0% | 1,971 | 586 | -70% | 0 | 0 | — |
case-20 | pass→pass | 10,528 | 9,630 | -9% | 1 | 1 | 0% | 2,128 | 2,170 | +2% | 0 | 0 | — |
case-21 | fail→fail | 12,612 | 7,567 | -40% | 1 | 1 | 0% | 2,090 | 1,396 | -33% | 0 | 0 | — |
case-22 | fail→pass | 9,158 | 3,230 | -65% | 1 | 1 | 0% | 1,706 | 751 | -56% | 0 | 0 | — |
case-23 | pass→pass | 8,506 | 9,056 | +6% | 1 | 1 | 0% | 1,871 | 1,800 | -4% | 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 +43 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are 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.