Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generate a conventional commit message for staged changes
.claude/skills/kunanonj-commit/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | -1% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | -13% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | 43% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -14% | 0% |
| case-06 | ✓→✓ | ＝ Same ✓ | 0% | 0% |
Generate a conventional commit message for staged changes.
git diff --cached to see staged changes<type>(<scope>): <subject>
[optional body]
[optional footer]feat: New featurefix: Bug fixdocs: Documentation onlystyle: Formatting, missing semicolons, etc.refactor: Code change that neither fixes nor adds featureperf: Performance improvementtest: Adding missing testschore: Maintenance tasksfeat(auth): add password reset functionality
Implement password reset flow with email verification.
Users can now request a reset link and set new password.
Closes #123fix(api): prevent race condition in order processing
Add mutex lock to ensure orders are processed sequentially.
This fixes duplicate charge issues reported by users.
Fixes #456refactor(cart): extract pricing logic to separate module
No functional changes. Improves testability and
separates concerns for future discount feature.After analyzing staged changes, suggest a commit message. Ask for confirmation before executing git commit -m "...".
$ARGUMENTS
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 1,818 | 1,661 | -9% | 1 | 1 | 0% | 205 | 478 | +133% | 0 | 0 | — |
case-02 | fail→fail | 2,437 | 3,095 | +27% | 1 | 1 | 0% | 495 | 456 | -8% | 0 | 0 | — |
case-03 | pass→pass | 3,976 | 2,722 | -32% | 1 | 1 | 0% | 799 | 694 | -13% | 0 | 0 | — |
case-04 | pass→pass | 4,128 | 3,204 | -22% | 1 | 1 | 0% | 624 | 891 | +43% | 0 | 0 | — |
case-05 | pass→pass | 4,425 | 2,226 | -50% | 1 | 1 | 0% | 882 | 759 | -14% | 0 | 0 | — |
case-06 | pass→pass | 4,305 | 2,597 | -40% | 1 | 1 | 0% | 812 | 816 | +0% | 0 | 0 | — |
case-07 | fail→pass | 6,213 | 4,194 | -32% | 1 | 1 | 0% | 1,115 | 1,107 | -1% | 0 | 0 | — |
case-08 | pass→pass | 3,806 | 1,985 | -48% | 1 | 1 | 0% | 751 | 673 | -10% | 0 | 0 | — |
case-09 | pass→pass | 4,585 | 2,693 | -41% | 1 | 1 | 0% | 872 | 820 | -6% | 0 | 0 | — |
case-10 | pass→pass | 3,218 | 2,001 | -38% | 1 | 1 | 0% | 651 | 728 | +12% | 0 | 0 | — |
case-11 | pass→pass | 5,912 | 4,184 | -29% | 1 | 1 | 0% | 1,198 | 1,167 | -3% | 0 | 0 | — |
case-12 | pass→pass | 3,899 | 2,601 | -33% | 1 | 1 | 0% | 770 | 703 | -9% | 0 | 0 | — |
case-13 | pass→pass | 5,327 | 4,556 | -14% | 1 | 1 | 0% | 1,061 | 1,066 | +0% | 0 | 0 | — |
case-14 | pass→pass | 9,576 | 5,514 | -42% | 1 | 1 | 0% | 1,610 | 1,241 | -23% | 0 | 0 | — |
case-15 | pass→pass | 8,468 | 4,648 | -45% | 1 | 1 | 0% | 1,625 | 1,185 | -27% | 0 | 0 | — |
case-16 | pass→pass | 6,162 | 3,266 | -47% | 1 | 1 | 0% | 1,219 | 839 | -31% | 0 | 0 | — |
case-17 | fail→fail | 4,154 | 4,038 | -3% | 1 | 1 | 0% | 735 | 561 | -24% | 0 | 0 | — |
case-18 | pass→pass | 3,673 | 3,125 | -15% | 1 | 1 | 0% | 549 | 900 | +64% | 0 | 0 | — |
case-19 | pass→pass | 2,260 | 1,644 | -27% | 1 | 1 | 0% | 386 | 587 | +52% | 0 | 0 | — |
case-20 | pass→pass | 9,664 | 5,460 | -44% | 1 | 1 | 0% | 1,627 | 1,263 | -22% | 0 | 0 | — |
case-21 | pass→pass | 6,585 | 3,289 | -50% | 1 | 1 | 0% | 1,360 | 989 | -27% | 0 | 0 | — |
case-22 | pass→pass | 10,962 | 5,275 | -52% | 1 | 1 | 0% | 1,885 | 1,454 | -23% | 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, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +5 percentage points is the difference between those two pass rates over the 20 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.