Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Command reference with practical examples from the tldr-pages project. Zero barrier — use with x-cmd, curl, or any tldr client. Use for "tldr", "cheatsheet", "command help", "examples", "how to".
.claude/skills/x-cmd-tldr/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -21% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -77% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -50% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -70% | 0% |
| case-17 | ✗→✓ | ▲ Improved | -79% | 0% |
Quick command reference with real-world examples. Better than man pages for "how do I use this command?"
bash# With x-cmd x tldr git # Git usage examples x tldr --lang zh python # Chinese docs x tldr --fz # Interactive search # Without x-cmd — use curl curl -s "https://raw.githubusercontent.com/tldr-pages/tldr/main/pages/common/git.md" # Or install a tldr client pip install tldr npm install -g tldr tldr tar
x tldr <cmd> — Show examples for a commandx tldr --lang zh <cmd> — Chinese languagex tldr --fz — Interactive fzf searchx tldr --cat <cmd> — Raw markdown displayx tldr --ls — List all available commandsx tldr --update — Update page cachepip install tldrnpm install -g tldrStarting with the essentials. Will add:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 4,870 | 6,887 | +41% | 1 | 1 | 0% | 753 | 596 | -21% | 0 | 0 | — |
case-02 | fail→pass | 20,710 | 2,799 | -86% | 1 | 1 | 0% | 3,355 | 756 | -77% | 0 | 0 | — |
case-03 | pass→pass | 19,173 | 4,427 | -77% | 1 | 1 | 0% | 3,574 | 916 | -74% | 0 | 0 | — |
case-04 | fail→pass | 10,162 | 4,414 | -57% | 1 | 1 | 0% | 1,569 | 780 | -50% | 0 | 0 | — |
case-05 | pass→pass | 7,421 | 3,838 | -48% | 1 | 1 | 0% | 1,271 | 733 | -42% | 0 | 0 | — |
case-06 | pass→pass | 5,507 | 4,931 | -10% | 1 | 1 | 0% | 974 | 1,158 | +19% | 0 | 0 | — |
case-07 | pass→pass | 3,283 | 4,712 | +44% | 1 | 1 | 0% | 513 | 872 | +70% | 0 | 0 | — |
case-08 | pass→pass | 4,013 | 3,114 | -22% | 1 | 1 | 0% | 579 | 723 | +25% | 0 | 0 | — |
case-09 | pass→pass | 8,242 | 6,548 | -21% | 1 | 1 | 0% | 1,320 | 1,226 | -7% | 0 | 0 | — |
case-10 | pass→pass | 4,179 | 3,487 | -17% | 1 | 1 | 0% | 623 | 860 | +38% | 0 | 0 | — |
case-11 | pass→pass | 15,522 | 3,451 | -78% | 1 | 1 | 0% | 2,515 | 664 | -74% | 0 | 0 | — |
case-12 | pass→pass | 8,469 | 3,409 | -60% | 1 | 1 | 0% | 971 | 749 | -23% | 0 | 0 | — |
case-13 | pass→pass | 4,072 | 3,110 | -24% | 1 | 1 | 0% | 581 | 788 | +36% | 0 | 0 | — |
case-14 | pass→pass | 5,918 | 2,786 | -53% | 1 | 1 | 0% | 919 | 692 | -25% | 0 | 0 | — |
case-15 | fail→pass | 24,128 | 2,831 | -88% | 1 | 1 | 0% | 2,431 | 728 | -70% | 0 | 0 | — |
case-16 | pass→fail | 5,481 | 3,279 | -40% | 1 | 1 | 0% | 837 | 876 | +5% | 0 | 0 | — |
case-17 | fail→pass | 18,747 | 3,031 | -84% | 1 | 1 | 0% | 3,187 | 676 | -79% | 0 | 0 | — |
case-18 | pass→pass | 12,761 | 2,437 | -81% | 1 | 1 | 0% | 1,958 | 666 | -66% | 0 | 0 | — |
case-19 | pass→pass | 14,734 | 3,047 | -79% | 1 | 1 | 0% | 2,455 | 740 | -70% | 0 | 0 | — |
case-20 | pass→pass | 16,771 | 16,186 | -3% | 1 | 1 | 0% | 2,966 | 825 | -72% | 0 | 0 | — |
case-21 | pass→pass | 4,718 | 5,368 | +14% | 1 | 1 | 0% | 799 | 1,096 | +37% | 0 | 0 | — |
case-22 | pass→pass | 5,295 | 4,951 | -6% | 1 | 1 | 0% | 864 | 1,086 | +26% | 0 | 0 | — |
case-23 | pass→pass | 6,109 | 7,464 | +22% | 1 | 1 | 0% | 971 | 1,094 | +13% | 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. 1 case got worse with the skill loaded, and it is 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.