Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →High-performance CSV processing with xan CLI for large tabular datasets, streaming transformations, and low-memory pipelines.
.claude/skills/foryourhealth111-pixel-xan/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -66% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -63% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -55% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -50% | 0% |
Use this skill when tabular work involves large CSV-like files and shell-oriented data pipelines.
filter, sort, dedup, groupby, frequency, join..xlsx formula/layout preservation) — use xlsx or excel-analysis.data-ml skills.scripts/xan.ps1powershellpowershell -ExecutionPolicy Bypass -File "$env:USERPROFILE\.codex\skills\xan\scripts\xan.ps1" count data.csv powershell -ExecutionPolicy Bypass -File "$env:USERPROFILE\.codex\skills\xan\scripts\xan.ps1" filter "score > 0.8" data.csv powershell -ExecutionPolicy Bypass -File "$env:USERPROFILE\.codex\skills\xan\scripts\xan.ps1" groupby category "sum(amount) as total" sales.csv
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | fail→pass | 12,645 | 8,140 | -36% | 1 | 1 | 0% | 2,226 | 1,120 | -50% | 0 | 0 | — |
case-01 | fail→pass | 13,545 | 3,195 | -76% | 1 | 1 | 0% | 2,529 | 867 | -66% | 0 | 0 | — |
case-02 | fail→pass | 11,939 | 2,968 | -75% | 1 | 1 | 0% | 2,139 | 782 | -63% | 0 | 0 | — |
case-03 | fail→pass | 9,501 | 2,410 | -75% | 1 | 1 | 0% | 1,635 | 730 | -55% | 0 | 0 | — |
case-04 | fail→pass | 8,370 | 2,450 | -71% | 1 | 1 | 0% | 1,444 | 722 | -50% | 0 | 0 | — |
case-05 | fail→pass | 12,332 | 3,927 | -68% | 1 | 1 | 0% | 2,113 | 909 | -57% | 0 | 0 | — |
case-06 | fail→pass | 10,350 | 2,635 | -75% | 1 | 1 | 0% | 1,846 | 791 | -57% | 0 | 0 | — |
case-07 | fail→pass | 12,484 | 3,141 | -75% | 1 | 1 | 0% | 2,259 | 825 | -63% | 0 | 0 | — |
case-08 | fail→pass | 11,039 | 4,036 | -63% | 1 | 1 | 0% | 2,102 | 1,005 | -52% | 0 | 0 | — |
case-10 | fail→pass | 12,269 | 4,117 | -66% | 1 | 1 | 0% | 2,161 | 1,083 | -50% | 0 | 0 | — |
case-11 | fail→pass | 7,992 | 2,933 | -63% | 1 | 1 | 0% | 1,344 | 779 | -42% | 0 | 0 | — |
case-12 | fail→pass | 22,245 | 2,216 | -90% | 1 | 1 | 0% | 1,903 | 729 | -62% | 0 | 0 | — |
case-13 | fail→pass | 15,448 | 3,166 | -80% | 1 | 1 | 0% | 2,723 | 775 | -72% | 0 | 0 | — |
case-14 | fail→pass | 10,048 | 2,577 | -74% | 1 | 1 | 0% | 1,796 | 744 | -59% | 0 | 0 | — |
case-15 | fail→pass | 7,365 | 2,292 | -69% | 1 | 1 | 0% | 1,422 | 684 | -52% | 0 | 0 | — |
case-16 | fail→pass | 11,208 | 3,183 | -72% | 1 | 1 | 0% | 1,837 | 830 | -55% | 0 | 0 | — |
case-17 | fail→pass | 9,950 | 2,804 | -72% | 1 | 1 | 0% | 1,569 | 819 | -48% | 0 | 0 | — |
case-18 | fail→pass | 10,039 | 2,685 | -73% | 1 | 1 | 0% | 1,791 | 760 | -58% | 0 | 0 | — |
case-19 | fail→pass | 7,991 | 2,234 | -72% | 1 | 1 | 0% | 1,369 | 620 | -55% | 0 | 0 | — |
case-20 | pass→pass | 10,579 | 5,477 | -48% | 1 | 1 | 0% | 1,744 | 1,353 | -22% | 0 | 0 | — |
case-21 | pass→pass | 6,716 | 3,377 | -50% | 1 | 1 | 0% | 1,245 | 929 | -25% | 0 | 0 | — |
case-22 | pass→pass | 21,635 | 6,753 | -69% | 1 | 1 | 0% | 2,516 | 1,473 | -41% | 0 | 0 | — |
case-23 | fail→pass | 12,578 | 2,492 | -80% | 1 | 1 | 0% | 2,072 | 744 | -64% | 0 | 0 | — |
case-24 | fail→pass | 7,676 | 2,362 | -69% | 1 | 1 | 0% | 1,499 | 798 | -47% | 0 | 0 | — |
case-25 | fail→pass | 10,053 | 2,591 | -74% | 1 | 1 | 0% | 1,880 | 789 | -58% | 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. 25 cases were attempted. The headline lift of +88 percentage points is the difference between those two pass rates over the 25 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.