Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Creates conventional git commits using git-agent and pushes to the remote repository. This skill should be used when the user asks to "commit and push", "push my changes", or wants to commit and immediately push to remote. The executing AI auto-derives its own co-author string from its runtime model identity (e.g., `Claude Opus 4.7 <noreply@anthropic.com>`, `Grok 4.5 <noreply@x.ai>`, `GLM-4.5 <noreply@zhipuai.cn>`) and passes it to `--co-author`. `$ARGUMENTS`, if provided, overrides the auto-der
.claude/skills/fradser-commit-and-push/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-20 | ✓→✗ | ▼ Worse | -77% | 0% |
| case-04 | ✓→✓ | ＝ Same ✓ | -36% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | -37% | 0% |
| case-07 | ✓→✓ | ＝ Same ✓ | -48% | 0% |
Create clean, atomic Conventional Commits using standard git commands and push them to origin.
bash git status --porcelain git diff --staged git diff
bash git add <file1> <file2> ...
Formulate a Conventional Commit message and commit: bash git commit -m "<type>(<scope>): <summary>"
Detect current branch and push: bash BRANCH=$(git branch --show-current) git push origin "$BRANCH" If pushing a new branch for the first time, append -u: bash git push -u origin "$BRANCH"
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,266 | 15,534 | +195% | 1 | 1 | 0% | 901 | 377 | -58% | 0 | 0 | — |
case-02 | fail→fail | 3,381 | 5,599 | +66% | 1 | 1 | 0% | 338 | 358 | +6% | 0 | 0 | — |
case-03 | fail→fail | 5,082 | 4,905 | -3% | 1 | 1 | 0% | 293 | 452 | +54% | 0 | 0 | — |
case-04 | pass→pass | 6,377 | 2,960 | -54% | 1 | 1 | 0% | 1,078 | 690 | -36% | 0 | 0 | — |
case-05 | pass→pass | 6,281 | 2,854 | -55% | 1 | 1 | 0% | 1,120 | 704 | -37% | 0 | 0 | — |
case-06 | fail→pass | 7,343 | 2,863 | -61% | 1 | 1 | 0% | 1,096 | 718 | -34% | 0 | 0 | — |
case-07 | pass→pass | 5,110 | 1,649 | -68% | 1 | 1 | 0% | 859 | 450 | -48% | 0 | 0 | — |
case-08 | pass→pass | 5,351 | 2,322 | -57% | 1 | 1 | 0% | 779 | 535 | -31% | 0 | 0 | — |
case-09 | pass→pass | 6,968 | 3,441 | -51% | 1 | 1 | 0% | 1,033 | 811 | -21% | 0 | 0 | — |
case-10 | pass→pass | 6,522 | 3,037 | -53% | 1 | 1 | 0% | 1,042 | 590 | -43% | 0 | 0 | — |
case-11 | pass→pass | 6,320 | 5,470 | -13% | 1 | 1 | 0% | 948 | 948 | 0% | 0 | 0 | — |
case-12 | pass→pass | 4,213 | 3,520 | -16% | 1 | 1 | 0% | 673 | 743 | +10% | 0 | 0 | — |
case-13 | pass→pass | 6,096 | 2,796 | -54% | 1 | 1 | 0% | 1,153 | 667 | -42% | 0 | 0 | — |
case-14 | pass→pass | 4,756 | 2,627 | -45% | 1 | 1 | 0% | 774 | 526 | -32% | 0 | 0 | — |
case-15 | pass→pass | 11,694 | 4,512 | -61% | 1 | 1 | 0% | 1,855 | 868 | -53% | 0 | 0 | — |
case-16 | pass→pass | 7,172 | 3,882 | -46% | 1 | 1 | 0% | 1,188 | 806 | -32% | 0 | 0 | — |
case-17 | pass→pass | 6,715 | 4,387 | -35% | 1 | 1 | 0% | 1,159 | 890 | -23% | 0 | 0 | — |
case-18 | pass→pass | 8,908 | 5,701 | -36% | 1 | 1 | 0% | 1,424 | 1,083 | -24% | 0 | 0 | — |
case-19 | pass→pass | 5,880 | 3,854 | -34% | 1 | 1 | 0% | 924 | 781 | -15% | 0 | 0 | — |
case-20 | pass→fail | 10,604 | 2,725 | -74% | 1 | 1 | 0% | 1,964 | 454 | -77% | 0 | 0 | — |
case-21 | pass→pass | 8,303 | 9,074 | +9% | 1 | 1 | 0% | 1,598 | 1,990 | +25% | 0 | 0 | — |
case-22 | pass→pass | 7,138 | 5,310 | -26% | 1 | 1 | 0% | 1,205 | 1,068 | -11% | 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 19 counted toward the lift figure. The other 3 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 0 percentage points is the difference between those two pass rates over the 19 comparable cases. 2 cases got worse with the skill loaded, and they are 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.