Install any skill in seconds. Free to start, no credit card required.
Get Started Free →**默认用 `-prune`,不用 filter。** Filter 让 find 进入目录;prune 直接跳过。
.claude/skills/x-cmd-find-prune/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -7% | 0% |
| case-03 | ✓→✓ | = Same ✓ | -1% | 0% |
默认用 -prune,不用 filter。 Filter 让 find 进入目录;prune 直接跳过。
shfind ROOT \( -name DIR1 -prune -o -name DIR2 -prune -o ... \) -o -print
末尾 -o -print(或其他动作)处理未匹配分支。
shnode_modules .git .svn .hg dist build out target vendor Pods DerivedData __pycache__ .venv .gradle .mvn .cache .tmp .DS_Store
隐藏目录:-name '.*' ! -name . -prune
shfind . \( -path '*/node_modules' -prune -o -path '*/.git' -prune \) -o -print
sh# 跳过噪音,找 .ts 文件 find . \( -name node_modules -prune -o -name .git -prune \) -o -type f -name '*.ts' -print # 跳过噪音,最近 7 天修改的文件 find . \( -name node_modules -prune \) -o -type f -mtime -7 -print
用户问".git/HEAD 里有什么",而搜索会 prune 掉 .git:
sh# 直接定位 find .git -name '*.txt' -print # 或从 prune 列表里去掉 .git find . \( -name node_modules -prune \) -o -print
-name '*.js',不是 -name *.js(shell 会展开)-prune:-name X -prune,不是只写 -name X-name vs -path:-name 匹配 basename,-path 匹配路径-o 短路:顺序重要,频率高的放前面shfind . -maxdepth 3 \ \( -name node_modules -prune -o \ -name .git -prune -o \ -name dist -prune -o \ -name build -prune -o \ -name __pycache__ -prune -o \ -name '.*' ! -name . -prune \) \ -o -print
x find — 自动排除常见噪音,默认 prunex bfind — BFS 逐层搜索(浅层优先)man find — 完整参考| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 7,755 | 4,876 | -37% | 1 | 1 | 0% | 1,327 | 1,469 | +11% | 0 | 0 | — |
case-02 | fail→pass | 9,406 | 6,511 | -31% | 1 | 1 | 0% | 1,511 | 1,199 | -21% | 0 | 0 | — |
case-03 | pass→pass | 10,572 | 7,437 | -30% | 1 | 1 | 0% | 1,643 | 1,621 | -1% | 0 | 0 | — |
case-04 | pass→pass | 15,785 | 9,956 | -37% | 1 | 1 | 0% | 2,672 | 2,210 | -17% | 0 | 0 | — |
case-05 | pass→pass | 14,482 | 8,135 | -44% | 1 | 1 | 0% | 2,356 | 1,791 | -24% | 0 | 0 | — |
case-06 | pass→pass | 7,195 | 4,717 | -34% | 1 | 1 | 0% | 1,004 | 1,235 | +23% | 0 | 0 | — |
case-07 | pass→pass | 7,851 | 5,218 | -34% | 1 | 1 | 0% | 1,266 | 1,438 | +14% | 0 | 0 | — |
case-08 | pass→pass | 6,965 | 5,979 | -14% | 1 | 1 | 0% | 929 | 1,295 | +39% | 0 | 0 | — |
case-09 | pass→pass | 14,383 | 11,029 | -23% | 1 | 1 | 0% | 2,246 | 2,479 | +10% | 0 | 0 | — |
case-10 | pass→pass | 29,191 | 14,672 | -50% | 1 | 1 | 0% | 2,335 | 2,759 | +18% | 0 | 0 | — |
case-11 | fail→pass | 13,525 | 11,550 | -15% | 1 | 1 | 0% | 2,021 | 2,647 | +31% | 0 | 0 | — |
case-12 | pass→pass | 9,733 | 8,494 | -13% | 1 | 1 | 0% | 1,563 | 2,020 | +29% | 0 | 0 | — |
case-13 | pass→pass | 9,087 | 3,965 | -56% | 1 | 1 | 0% | 1,336 | 1,241 | -7% | 0 | 0 | — |
case-14 | pass→pass | 8,385 | 5,243 | -37% | 1 | 1 | 0% | 1,239 | 1,194 | -4% | 0 | 0 | — |
case-15 | fail→pass | 17,499 | 4,768 | -73% | 1 | 1 | 0% | 1,453 | 1,349 | -7% | 0 | 0 | — |
case-16 | pass→pass | 30,017 | 11,859 | -60% | 1 | 1 | 0% | 2,586 | 2,659 | +3% | 0 | 0 | — |
case-17 | pass→pass | 6,964 | 6,916 | -1% | 1 | 1 | 0% | 1,031 | 1,666 | +62% | 0 | 0 | — |
case-18 | pass→pass | 10,569 | 8,843 | -16% | 1 | 1 | 0% | 1,539 | 2,123 | +38% | 0 | 0 | — |
case-19 | pass→pass | 9,160 | 21,547 | +135% | 1 | 1 | 0% | 1,416 | 1,804 | +27% | 0 | 0 | — |
case-20 | pass→pass | 10,534 | 9,182 | -13% | 1 | 1 | 0% | 1,837 | 1,260 | -31% | 0 | 0 | — |
case-21 | pass→pass | 20,037 | 38,080 | +90% | 1 | 1 | 0% | 3,341 | 4,227 | +27% | 0 | 0 | — |
case-22 | pass→pass | 11,207 | 10,233 | -9% | 1 | 1 | 0% | 1,908 | 2,477 | +30% | 0 | 0 | — |
case-23 | pass→pass | 8,584 | 4,287 | -50% | 1 | 1 | 0% | 1,198 | 1,199 | +0% | 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. The headline lift of +17 percentage points is the difference between those two pass rates over the 23 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.