git-flow skill
Use when creating a branch, using worktrees for parallel work, committing, or wrapping up. Covers branch naming, conventional commits, short-lived branches, worktree isolation, and final merge/cleanup.
- GitHub 스타
- 25
- 파일
- 1
- 마지막 커밋
- 2026년 9월 15일
- 라이선스
- MIT
- 허용된 도구
- —
git-flow skill은 무엇을 하나요?
git-flow은(는) agent skill입니다. Use when creating a branch, using worktrees for parallel work, committing, or wrapping up. Covers branch naming, conventional commits, short-lived branches, worktree isolation, and final merge/cleanup.
git-flow skill 설치 방법
각 명령은 skill 폴더를 클라이언트의 skills 디렉터리에 복사합니다. 클라이언트를 재시작하거나 새 세션을 시작하여 로드하세요.플러그인에 git-flow skill 설치
/plugin marketplace add chipfighter/coding-discipline /plugin install coding-discipline@coding-discipline
Claude Code에 git-flow skill 설치
~/.claude/skills/git-flow/SKILL.md
$ npx degit chipfighter/coding-discipline/plugins/coding-discipline/skills/git-flow ~/.claude/skills/git-flowClaude Code (project)에 git-flow skill 설치
.claude/skills/git-flow/SKILL.md
$ npx degit chipfighter/coding-discipline/plugins/coding-discipline/skills/git-flow .claude/skills/git-flowCodex에 git-flow skill 설치
~/.agents/skills/git-flow/SKILL.md
$ npx degit chipfighter/coding-discipline/plugins/coding-discipline/skills/git-flow ~/.agents/skills/git-flowGemini CLI에 git-flow skill 설치
~/.gemini/skills/git-flow/SKILL.md
$ npx degit chipfighter/coding-discipline/plugins/coding-discipline/skills/git-flow ~/.gemini/skills/git-flowClaude.ai에 git-flow skill 설치
npx degit chipfighter/coding-discipline/plugins/coding-discipline/skills/git-flow git-flow && zip -r git-flow.zip git-flow # claude.ai → Customize → Skills → + → Create skill → Upload a skill → git-flow.zip
SKILL.md
GitHub에서 보기Branch naming rules, protected branches, whether to merge directly or through a PR / MR, whether merging triggers CI, and how to choose versions / who creates tags are project-specific rules governed by this repository's
AGENTS.md/CLAUDE.md. This skill provides only general discipline.
Version / tag closeout (only for projects that manage versions with tags + CHANGELOG; a tag is a version bookmark—whether it also means release / deployment is defined by the project guide document)
- Choose the version before creating the branch: first confirm the intended tag with the owner—state the previous tag, what remains in the backlog, and whether this work is a feature or a fix, then recommend a version. Wait for the owner's confirmation before creating the branch; do not finish the work only to discover there is no agreed release target.
- Follow SemVer: major = breaking change, minor = backward-compatible feature, patch = backward-compatible fix. For 0.x, use the looser convention (minor = capability, patch = fix), and when uncertain prefer the larger bump. The project guide document defines what counts as a minor change.
- Freeze once decided: after the target version is confirmed, freeze its scope. New requirements arising midway go to the backlog / next version by default. Including one in the current version requires explicit owner approval and a clear statement of what it displaces or defers.
- Use a component, not a version number, as the commit scope (
feat(auth):, notfeat(v1.2):). Branches are deleted; commit history is not. The history remains accurate if the version changes. - Version closeout sequence: complete the scope → merge the branch into the baseline → get CI green → move the CHANGELOG entry from pending into the new version → create the tag (who creates it and how are project-specific).
Branches and commits
- Create a new branch for each development effort. Never edit directly on a protected main branch or someone else's branch. If the user or client has explicitly provided a working branch, keep using it instead of creating a pointless one. Keep branches short-lived: integrate them quickly rather than letting them drift.
- Use a type + short description in branch names (for example,
feat/search-paginationorfix/login-timeout); follow the project's naming convention when it has one. - Use conventional commits:
type(scope): description(feat/fix/refactor/docs/chore…). Commit in small steps, each independently reversible.
Parallel worktrees (when running several workstreams at once or isolating work from the current workspace)
- First check whether you are already in an isolated workspace. If so, do not nest another one.
- Prefer built-in worktree management in clients such as Claude Code / Codex. Do not manually run
git worktree addand create a workspace the client cannot see; fall back to Git only when the client lacks this capability. - Before creating a project-local worktree under
.worktrees/, confirm that the directory is gitignored; otherwise its contents may be committed accidentally. - After creating it, install dependencies and run the baseline tests. Report existing failures before starting work.
Wrap-up
- Verify that all tests pass before wrapping up (see verify-before-done). A failing branch must not proceed to merge / PR.
- Offer explicit choices instead of asking an open-ended “What next?”: ① merge into the baseline branch ② push and open a PR/MR ③ keep as-is ④ discard.
- Always in this order: merge and verify success → remove the worktree → delete the branch. A branch still referenced by a worktree cannot be deleted; before running
git worktree remove,cdback to the main repository. - Clean up only worktrees you created; do not touch client-created ones. Discarding requires typed confirmation. Do not force-push unless explicitly asked.
이 스킬의 파일
1git-flow skill 질문
Claude Code에 git-flow skill을 어떻게 설치하나요?
npx degit chipfighter/coding-discipline/plugins/coding-discipline/skills/git-flow ~/.claude/skills/git-flow를 실행합니다. Claude Code는 ~/.claude/skills(모든 프로젝트) 또는 .claude/skills(한 프로젝트)에서 skills를 로드하며, 설명이 작업과 일치할 때 skill을 사용합니다.
git-flow skill을 사용할 수 있는 클라이언트는 무엇인가요?
Agent Skills 형식을 읽는 모든 클라이언트(Claude Code, Codex, Gemini CLI, claude.ai(ZIP으로 업로드))가 사용할 수 있습니다. skill 내 스크립트는 클라이언트가 코드 실행을 허용하는 경우에만 실행됩니다.