Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Generates meaningful, conventional commit messages automatically. Use when committing code changes. Analyzes diff to create descriptive commits following best practices.
.claude/skills/majiayu000-git-commit-smart/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-09 | ✗→✓ | ▲ Improved | 160% | 0% |
| case-05 | ✓→✗ | ▼ Worse | 178% | 0% |
| case-01 | ✓→✓ | ＝ Same ✓ | 117% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 90% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 138% | 0% |
> Inspired by claude-code-plugins-plus
Automatically generate meaningful commit messages by analyzing staged changes, following conventional commit format.
<type>(<scope>): <subject>
<body>
<footer>| Type | Description | |------|-------------| | feat | New feature | | fix | Bug fix | | docs | Documentation only | | style | Formatting, no code change | | refactor | Code change, no new feature or fix | | perf | Performance improvement | | test | Adding/updating tests | | chore | Maintenance tasks | | ci | CI/CD changes | | build | Build system changes |
Optional, indicates the section of codebase:
auth, api, ui, db, config, etc.BREAKING CHANGE: descriptionCloses #123, Fixes #456bash# View staged changes git diff --staged # View changed files git diff --staged --name-only # View stats git diff --staged --stat
Based on changes:
featfixdocstestchore or buildLook at file paths:
src/auth/* → scope: authsrc/api/* → scope: apitests/* → scope: testSummarize the change:
feat(auth): add OAuth2 login support
fix(api): handle null response in user endpoint
docs: update README with installation steps
refactor(db): extract query builder to separate moduleFor complex changes:
feat(auth): add OAuth2 login support
Implement OAuth2 authentication flow with support for
Google and GitHub providers. This allows users to sign
in without creating a separate account.
- Add OAuth2 client configuration
- Implement callback handlers
- Store OAuth tokens securely
- Add provider selection UIfeat(ui): add dark mode togglefix(api): prevent duplicate user creation
Check for existing email before inserting new user record.
Race condition could cause duplicate entries when requests
arrived simultaneously.
Fixes #234feat(api)!: change response format to JSON:API
BREAKING CHANGE: All API responses now follow JSON:API
specification. Clients must update their parsers.
Migration guide: docs/migration-v2.mdrefactor: extract validation logic to shared module
Move duplicate validation code from controllers to
a shared validation module. Reduces code duplication
and ensures consistent validation across endpoints.❌ "fix stuff"
❌ "update code"
❌ "WIP"
❌ "changes"
❌ "fixed bug"
❌ "misc updates"
❌ Commits with 50+ files and vague messagegit diff --staged alwaysbash# Stage specific files git add <file> # Stage all changes git add -A # Interactive staging git add -p # Commit with message git commit -m "type(scope): subject" # Commit with body git commit -m "type(scope): subject" -m "body paragraph" # Amend last commit git commit --amend
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 4,720 | 5,445 | +15% | 1 | 1 | 0% | 846 | 1,837 | +117% | 0 | 0 | — |
case-02 | pass→pass | 5,275 | 4,519 | -14% | 1 | 1 | 0% | 940 | 1,789 | +90% | 0 | 0 | — |
case-03 | pass→pass | 4,304 | 3,463 | -20% | 1 | 1 | 0% | 682 | 1,622 | +138% | 0 | 0 | — |
case-04 | pass→pass | 3,741 | 3,032 | -19% | 1 | 1 | 0% | 585 | 1,534 | +162% | 0 | 0 | — |
case-05 | pass→fail | 3,900 | 3,274 | -16% | 1 | 1 | 0% | 585 | 1,628 | +178% | 0 | 0 | — |
case-06 | pass→pass | 3,885 | 3,725 | -4% | 1 | 1 | 0% | 664 | 1,729 | +160% | 0 | 0 | — |
case-07 | pass→pass | 4,872 | 4,358 | -11% | 1 | 1 | 0% | 795 | 1,792 | +125% | 0 | 0 | — |
case-08 | pass→pass | 4,621 | 2,677 | -42% | 1 | 1 | 0% | 691 | 1,522 | +120% | 0 | 0 | — |
case-09 | fail→pass | 3,656 | 3,439 | -6% | 1 | 1 | 0% | 639 | 1,662 | +160% | 0 | 0 | — |
case-10 | pass→pass | 9,266 | 5,893 | -36% | 1 | 1 | 0% | 1,456 | 2,035 | +40% | 0 | 0 | — |
case-11 | pass→pass | 8,840 | 7,297 | -17% | 1 | 1 | 0% | 1,546 | 2,288 | +48% | 0 | 0 | — |
case-12 | pass→pass | 2,766 | 2,429 | -12% | 1 | 1 | 0% | 447 | 1,382 | +209% | 0 | 0 | — |
case-13 | pass→pass | 5,783 | 4,771 | -17% | 1 | 1 | 0% | 930 | 1,791 | +93% | 0 | 0 | — |
case-14 | pass→pass | 8,373 | 4,862 | -42% | 1 | 1 | 0% | 1,353 | 1,806 | +33% | 0 | 0 | — |
case-15 | pass→pass | 6,080 | 6,038 | -1% | 1 | 1 | 0% | 937 | 2,002 | +114% | 0 | 0 | — |
case-16 | pass→pass | 4,167 | 2,621 | -37% | 1 | 1 | 0% | 569 | 1,446 | +154% | 0 | 0 | — |
case-17 | pass→pass | 5,546 | 3,920 | -29% | 1 | 1 | 0% | 933 | 1,666 | +79% | 0 | 0 | — |
case-18 | pass→pass | 5,014 | 3,373 | -33% | 1 | 1 | 0% | 851 | 1,579 | +86% | 0 | 0 | — |
case-19 | pass→pass | 5,208 | 3,226 | -38% | 1 | 1 | 0% | 881 | 1,564 | +78% | 0 | 0 | — |
case-20 | pass→pass | 2,462 | 1,793 | -27% | 1 | 1 | 0% | 357 | 1,263 | +254% | 0 | 0 | — |
case-21 | pass→pass | 3,041 | 2,559 | -16% | 1 | 1 | 0% | 501 | 1,400 | +179% | 0 | 0 | — |
case-22 | pass→pass | 8,859 | 8,628 | -3% | 1 | 1 | 0% | 1,532 | 2,557 | +67% | 0 | 0 | — |
case-23 | pass→pass | 9,230 | 8,377 | -9% | 1 | 1 | 0% | 1,490 | 2,418 | +62% | 0 | 0 | — |
case-24 | pass→pass | 5,478 | 6,910 | +26% | 1 | 1 | 0% | 946 | 2,312 | +144% | 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. 24 cases were attempted. The headline lift of 0 percentage points is the difference between those two pass rates over the 24 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.