Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create a Conventional Commit message for staged changes and run git commit. Use when changes are staged and you need a commit message in the form type(scope): subject, with a <=100 character imperative header and optional 'ref AL-###' body line. Retained for reference; use only when explicitly asked for this commit workflow.
.claude/skills/asymmetric-al-commit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 8% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 6% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 19% | 0% |
Retained for reference only. This skill is not part of the active repo routing and should not be used unless a user explicitly asks for this commit workflow.
Name: commit Purpose: Create a Conventional Commit message for staged changes and run git commit. Use this skill when you are ready to commit work in this repo.
Applies when: Changes are staged and you need a commit message. Do not use when: Nothing is staged or you are not using Git.
feat, fix, docs, style, refactor, test, chore, perf, ci, build, revert.type(scope): subject (scope optional).ref AL-### in the body only when applicable.AL-### keys (branch, diff, or recent commits).git commit.git status shows staged changesref AL-### included only when applicablebashgit commit -m "fix(auth): handle empty session"
bashgit commit -m "feat(billing): add invoice export" -m "Add export endpoint and UI action" -m "ref AL-123"
ref AL-### when no issue applies| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,524 | 3,214 | -9% | 1 | 1 | 0% | 478 | 865 | +81% | 0 | 0 | — |
case-02 | pass→pass | 6,710 | 3,469 | -48% | 1 | 1 | 0% | 1,064 | 1,123 | +6% | 0 | 0 | — |
case-03 | fail→pass | 4,095 | 4,720 | +15% | 1 | 1 | 0% | 729 | 1,245 | +71% | 0 | 0 | — |
case-04 | pass→pass | 5,063 | 3,416 | -33% | 1 | 1 | 0% | 874 | 1,044 | +19% | 0 | 0 | — |
case-05 | pass→pass | 6,100 | 4,056 | -34% | 1 | 1 | 0% | 979 | 1,111 | +13% | 0 | 0 | — |
case-06 | pass→pass | 3,722 | 2,942 | -21% | 1 | 1 | 0% | 589 | 969 | +65% | 0 | 0 | — |
case-07 | fail→pass | 6,442 | 4,048 | -37% | 1 | 1 | 0% | 1,089 | 1,173 | +8% | 0 | 0 | — |
case-08 | pass→pass | 7,080 | 3,268 | -54% | 1 | 1 | 0% | 1,033 | 1,075 | +4% | 0 | 0 | — |
case-09 | pass→pass | 4,512 | 4,552 | +1% | 1 | 1 | 0% | 725 | 1,213 | +67% | 0 | 0 | — |
case-10 | pass→pass | 2,364 | 2,387 | +1% | 1 | 1 | 0% | 390 | 822 | +111% | 0 | 0 | — |
case-11 | pass→pass | 2,905 | 2,826 | -3% | 1 | 1 | 0% | 416 | 898 | +116% | 0 | 0 | — |
case-12 | pass→pass | 4,473 | 3,528 | -21% | 1 | 1 | 0% | 727 | 1,037 | +43% | 0 | 0 | — |
case-13 | pass→pass | 6,061 | 3,846 | -37% | 1 | 1 | 0% | 1,005 | 1,127 | +12% | 0 | 0 | — |
case-14 | pass→pass | 7,161 | 3,797 | -47% | 1 | 1 | 0% | 1,155 | 1,128 | -2% | 0 | 0 | — |
case-15 | pass→pass | 5,800 | 2,362 | -59% | 1 | 1 | 0% | 946 | 855 | -10% | 0 | 0 | — |
case-16 | pass→pass | 6,714 | 2,398 | -64% | 1 | 1 | 0% | 955 | 811 | -15% | 0 | 0 | — |
case-17 | pass→pass | 5,625 | 2,755 | -51% | 1 | 1 | 0% | 936 | 889 | -5% | 0 | 0 | — |
case-18 | pass→pass | 3,857 | 3,442 | -11% | 1 | 1 | 0% | 610 | 996 | +63% | 0 | 0 | — |
case-19 | pass→pass | 6,786 | 5,087 | -25% | 1 | 1 | 0% | 1,089 | 1,262 | +16% | 0 | 0 | — |
case-20 | pass→pass | 5,671 | 3,353 | -41% | 1 | 1 | 0% | 960 | 1,055 | +10% | 0 | 0 | — |
case-21 | fail→pass | 7,625 | 4,545 | -40% | 1 | 1 | 0% | 1,260 | 1,246 | -1% | 0 | 0 | — |
case-22 | pass→pass | 3,837 | 2,910 | -24% | 1 | 1 | 0% | 592 | 934 | +58% | 0 | 0 | — |
case-23 | pass→pass | 6,351 | 6,500 | +2% | 1 | 1 | 0% | 1,244 | 1,733 | +39% | 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 +13 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.