Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Normalize line endings for cross-platform file handling with CRLF/LF conversion and git configuration.
.claude/skills/a5c-ai-line-ending-normalizer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | -8% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -54% | 0% |
Normalize line endings for cross-platform compatibility.
typescriptexport type LineEnding = 'lf' | 'crlf' | 'mixed'; export function detectLineEnding(content: string): LineEnding { const crlf = (content.match(/\r\n/g) || []).length; const lf = (content.match(/(?<!\r)\n/g) || []).length; if (crlf > 0 && lf > 0) return 'mixed'; if (crlf > 0) return 'crlf'; return 'lf'; } export function normalizeLineEndings(content: string, target: 'lf' | 'crlf' = 'lf'): string { const normalized = content.replace(/\r\n/g, '\n').replace(/\r/g, '\n'); return target === 'crlf' ? normalized.replace(/\n/g, '\r\n') : normalized; } // .gitattributes content export const gitattributes = ` * text=auto eol=lf *.bat text eol=crlf *.cmd text eol=crlf *.ps1 text eol=crlf *.sh text eol=lf `;
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 11,823 | 11,896 | +1% | 1 | 1 | 0% | 2,716 | 2,952 | +9% | 0 | 0 | — |
case-02 | fail→pass | 11,196 | 8,883 | -21% | 1 | 1 | 0% | 2,324 | 2,145 | -8% | 0 | 0 | — |
case-03 | fail→pass | 18,053 | 12,382 | -31% | 1 | 1 | 0% | 3,120 | 3,143 | +1% | 0 | 0 | — |
case-04 | fail→fail | 18,395 | 10,068 | -45% | 1 | 1 | 0% | 3,655 | 2,454 | -33% | 0 | 0 | — |
case-05 | fail→pass | 12,075 | 10,339 | -14% | 1 | 1 | 0% | 2,412 | 2,425 | +1% | 0 | 0 | — |
case-06 | fail→fail | 8,904 | 9,173 | +3% | 1 | 1 | 0% | 1,691 | 2,240 | +32% | 0 | 0 | — |
case-07 | pass→pass | 12,408 | 8,940 | -28% | 1 | 1 | 0% | 2,213 | 1,987 | -10% | 0 | 0 | — |
case-08 | fail→pass | 11,633 | 11,070 | -5% | 1 | 1 | 0% | 2,172 | 2,406 | +11% | 0 | 0 | — |
case-09 | fail→fail | 8,559 | 8,694 | +2% | 1 | 1 | 0% | 1,681 | 2,000 | +19% | 0 | 0 | — |
case-10 | fail→fail | 10,136 | 8,366 | -17% | 1 | 1 | 0% | 1,765 | 2,059 | +17% | 0 | 0 | — |
case-11 | pass→pass | 10,331 | 5,285 | -49% | 1 | 1 | 0% | 1,808 | 1,305 | -28% | 0 | 0 | — |
case-12 | fail→fail | 3,515 | 4,264 | +21% | 1 | 1 | 0% | 635 | 1,126 | +77% | 0 | 0 | — |
case-13 | fail→pass | 9,298 | 2,103 | -77% | 1 | 1 | 0% | 1,573 | 720 | -54% | 0 | 0 | — |
case-14 | fail→pass | 8,058 | 5,302 | -34% | 1 | 1 | 0% | 1,502 | 1,480 | -1% | 0 | 0 | — |
case-15 | fail→fail | 9,317 | 5,393 | -42% | 1 | 1 | 0% | 1,623 | 1,269 | -22% | 0 | 0 | — |
case-16 | fail→fail | 12,855 | 6,878 | -46% | 1 | 1 | 0% | 2,364 | 1,781 | -25% | 0 | 0 | — |
case-17 | fail→fail | 5,179 | 4,121 | -20% | 1 | 1 | 0% | 879 | 1,080 | +23% | 0 | 0 | — |
case-18 | fail→pass | 9,301 | 3,051 | -67% | 1 | 1 | 0% | 1,561 | 1,013 | -35% | 0 | 0 | — |
case-19 | fail→fail | 8,195 | 4,201 | -49% | 1 | 1 | 0% | 1,311 | 1,208 | -8% | 0 | 0 | — |
case-20 | fail→fail | 14,922 | 10,058 | -33% | 1 | 1 | 0% | 2,745 | 2,501 | -9% | 0 | 0 | — |
case-21 | fail→fail | 11,872 | 10,583 | -11% | 1 | 1 | 0% | 2,219 | 2,426 | +9% | 0 | 0 | — |
case-22 | fail→fail | 11,826 | 11,436 | -3% | 1 | 1 | 0% | 2,221 | 2,812 | +27% | 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 +32 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.