Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Safely stage, commit, and push intended git changes with conventional commit messages. Use for ordinary non-release pushes when explicitly asked to push, save work remotely, or share a completed change.
.claude/skills/sickn33-git-pushing/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -20% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -31% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -35% | 0% |
Stage only intended changes, create a conventional commit, and push to the remote branch.
Automatically activate when the user:
Before staging, inspect git status --short --branch, confirm the intended files, and fetch the upstream branch when a concurrent push is plausible. Do not absorb unrelated dirty files.
Read repository policy before choosing the destination branch. If main or master is protected, or the repository defines a maintainer command such as merge:batch, create or use a topic branch and finish through the required pull-request checks. A user request to “push to main” describes the desired final state; it does not authorize bypassing server-side protection. Never keep retrying a direct push after a protected-branch rejection.
The helper requires an empty live index and a conventional commit message before it stages anything. It locks the live index, builds and validates the commit in an isolated temporary index, rejects -- without paths, and atomically updates the branch only if its parent is unchanged.
The helper honors branch.<name>.pushRemote, remote.pushDefault, and the branch's configured upstream, in that order. For a new branch without those settings, it requires origin and establishes origin/<branch>. It rejects detached HEAD and invalid remote configurations before staging.
Do not use this skill for a maintainer merge batch, canonical synchronization, versioned repository release, tag publication, or a repository with an explicit merge:batch, release:prepare, or release:publish workflow. Use that repository's maintainer/release flow instead; it owns pull-request evidence, protected-branch checks, generated files, tags, and publication verification.
Use the helper only after the safety gates pass. Resolve the installed directory that contains this SKILL.md and substitute its absolute path for <skill-directory> below; do not assume the current working directory is the catalog repository. With no paths the helper stages all current changes, so use that form only when every dirty file belongs to the requested commit:
bashbash "<skill-directory>/scripts/smart_commit.sh"
With custom message:
bashbash "<skill-directory>/scripts/smart_commit.sh" "feat: add feature"
To stage only named files, pass them after --:
bashbash "<skill-directory>/scripts/smart_commit.sh" "fix: scope change" -- path/to/file
The helper handles isolated staging, commit creation, and push; it does not replace validation, release tooling, or a rebase required by an advanced upstream branch.
files ref backend; it rejects reftable repositories before creating a commit because their refs cannot use the filesystem lock protocol.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 3,986 | 8,900 | +123% | 1 | 1 | 0% | 670 | 997 | +49% | 0 | 0 | — |
case-02 | fail→fail | 7,429 | 5,195 | -30% | 1 | 1 | 0% | 1,234 | 982 | -20% | 0 | 0 | — |
case-03 | fail→fail | 15,613 | 5,989 | -62% | 1 | 1 | 0% | 779 | 1,046 | +34% | 0 | 0 | — |
case-04 | fail→pass | 11,930 | 7,114 | -40% | 1 | 1 | 0% | 1,898 | 1,926 | +1% | 0 | 0 | — |
case-05 | pass→pass | 5,286 | 3,658 | -31% | 1 | 1 | 0% | 925 | 1,332 | +44% | 0 | 0 | — |
case-06 | pass→pass | 6,053 | 4,188 | -31% | 1 | 1 | 0% | 991 | 1,345 | +36% | 0 | 0 | — |
case-07 | pass→pass | 5,219 | 3,407 | -35% | 1 | 1 | 0% | 952 | 1,092 | +15% | 0 | 0 | — |
case-08 | fail→pass | 8,609 | 7,041 | -18% | 1 | 1 | 0% | 1,360 | 1,162 | -15% | 0 | 0 | — |
case-09 | pass→pass | 9,438 | 5,464 | -42% | 1 | 1 | 0% | 1,487 | 1,668 | +12% | 0 | 0 | — |
case-10 | fail→pass | 10,247 | 5,252 | -49% | 1 | 1 | 0% | 1,826 | 1,459 | -20% | 0 | 0 | — |
case-11 | pass→pass | 6,664 | 2,925 | -56% | 1 | 1 | 0% | 1,060 | 1,221 | +15% | 0 | 0 | — |
case-12 | fail→pass | 9,656 | 1,617 | -83% | 1 | 1 | 0% | 1,453 | 999 | -31% | 0 | 0 | — |
case-13 | fail→pass | 12,078 | 3,230 | -73% | 1 | 1 | 0% | 1,913 | 1,243 | -35% | 0 | 0 | — |
case-14 | pass→pass | 7,056 | 3,889 | -45% | 1 | 1 | 0% | 1,162 | 1,326 | +14% | 0 | 0 | — |
case-15 | pass→pass | 10,571 | 4,312 | -59% | 1 | 1 | 0% | 1,817 | 1,464 | -19% | 0 | 0 | — |
case-16 | pass→pass | 8,783 | 2,332 | -73% | 1 | 1 | 0% | 1,306 | 1,062 | -19% | 0 | 0 | — |
case-17 | fail→pass | 6,551 | 3,317 | -49% | 1 | 1 | 0% | 1,118 | 1,340 | +20% | 0 | 0 | — |
case-18 | pass→pass | 9,674 | 5,150 | -47% | 1 | 1 | 0% | 1,670 | 1,553 | -7% | 0 | 0 | — |
case-19 | pass→pass | 8,872 | 2,987 | -66% | 1 | 1 | 0% | 1,314 | 1,318 | +0% | 0 | 0 | — |
case-20 | pass→pass | 9,777 | 3,188 | -67% | 1 | 1 | 0% | 1,628 | 1,305 | -20% | 0 | 0 | — |
case-21 | pass→pass | 9,340 | 3,262 | -65% | 1 | 1 | 0% | 1,609 | 1,304 | -19% | 0 | 0 | — |
case-22 | pass→pass | 8,508 | 3,801 | -55% | 1 | 1 | 0% | 1,392 | 1,455 | +5% | 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 +27 percentage points is the difference between those two pass rates over the 19 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.