git-atomic-commit

Always produce small commits where each commit has one logical purpose and passes checks.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "git-atomic-commit" with this command: npx skills add chasebuild/agent-skills/chasebuild-agent-skills-git-atomic-commit

Git Atomic Commit

Always produce small commits where each commit has one logical purpose and passes checks.

Required Guardrails

  • Run git-run-checks before the first commit and after each meaningful commit group.

  • Never mix refactor + behavior change + formatting in one commit unless unavoidable.

  • Keep commit ordering dependency-safe (prerequisites first).

Atomic Slicing Process

  • Inspect changes with git status --short and git diff .

  • Draft commit slices by concern:

  • mechanical move/rename

  • API or schema changes

  • behavior changes

  • tests

  • docs/chore

  • Stage only one slice (git add -p or path-based staging).

  • Verify staged diff with git diff --cached .

  • Commit with a Conventional Commit message (delegate to git-conventional-commit ).

  • Repeat until working tree is clean.

Quality Checks per Slice

  • Commit compiles and tests at least at smoke level.

  • Message explains intent, not implementation trivia.

  • Follow-up commits do not silently depend on unstaged files.

Optional Helper Script

Use ./scripts/suggest-slices.sh to get candidate commit groupings from file paths.

Boris-Inspired Operating Rules

  • Read first, code less, and edit only what is necessary.

  • Keep context tight and decisions explicit.

  • Prefer deterministic command sequences over ad-hoc experimentation.

  • Surface assumptions and risks before finalizing each commit.

Claude Mapping (Optional)

  • Use parallel agents only for independent subproblems; merge results into one coherent commit plan.

  • Keep each subagent assigned to one commit slice to avoid overlap.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Automation

git-pr-workflow

No summary provided by upstream source.

Repository SourceNeeds Review
Automation

git-conventional-commit

No summary provided by upstream source.

Repository SourceNeeds Review
Automation

coding-guidelines

No summary provided by upstream source.

Repository SourceNeeds Review