Token导航 LogoToken导航TokenDH.com
前端设计需要联网github未标认证来源可访问许可证需确认审计通过

git-commitGit 提交

Agent Skill

git-commit 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

376

周安装

16

GitHub Stars

公开资料未说明

下载量

132
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:git-commit(Git 提交)
来源仓库:https://github.com/vamdawn/ai-forge
仓库路径:skills/git-commit
安装命令:
npx skills add https://github.com/vamdawn/ai-forge --skill git-commit
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/vamdawn/ai-forge --skill git-commit

简介

git-commit 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息。

  • 适合在 Codex、Claude、Cursor、Gemini CLI 中围绕仓库状态、代码变更或协作事项进行整理。
  • 通过 npx skills add 命令从指定仓库安装,需结合原始 README 确认具体用法。
  • 安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。
  • 当前暂无更多功能细节,可参考来源仓库进一步了解实现逻辑和使用示例。

SKILL.md

Git Commit

Create atomic, well-formatted commits with emoji conventional commit messages.

Workflow

Step 1: Assess

Run git status. Then check:

Abort if:

  • Unresolved merge conflicts exist → inform the user and STOP
  • No modified, staged, or untracked files exist → inform the user and STOP

Gather context:

  1. git log --oneline -5 — learn the project's existing commit style
  2. git diff --staged if files are already staged; otherwise git diff for tracked files
  3. For untracked files shown in git status, use Read to view their content

If files are already staged: work with ONLY those files. Do NOT stage additional files.

Step 2: Classify and Group

For each changed file, determine its type (feat/fix/refactor/docs/...) and concern (which subsystem or feature it belongs to).

Group into logical units — each one a single coherent purpose that stands alone as a git log entry. Apply keep-together rules first to form atomic units, then evaluate split rules between those units.

Split when ANY apply:

  1. Changes serve different purposes (bug fix + new feature)
  2. Changes touch unrelated subsystems (backend API + frontend CSS)
  3. Source code changes mixed with unrelated documentation
  4. Diff exceeds ~200 lines with separable concerns
  5. Formatting/style changes mixed with logic changes

Keep together:

  • Feature/fix + its tests
  • Code + its type definitions
  • Multi-file rename of the same symbol
  • Code + documentation describing that same code

Step 3: Stage

Single logical unitgit add <file1> <file2>..., proceed to Step 4.

Multiple logical units

  1. Present the plan: N logical groups: 1. [type]: [desc] (files) 2....
  2. STOP and wait for the user to confirm — do not proceed until they respond
  3. For each group (foundational changes first): stage → commit (Step 4) → loop back

Rules:

  • ALWAYS use explicit file paths — NEVER git add -A or git add.
  • NEVER stage secret files (.env, credentials, keys, tokens) — warn the user and skip

Step 4: Commit

Format: <emoji> <type>: <description>

Rules:

  • Imperative mood, present tense ("add" not "added")
  • Subject line under 72 characters
  • Lowercase after colon ("feat: add..." not "feat: Add...")
  • No co-authorship footer
  • Match project's commit style observed in Step 1
TypeEmojiWhen to use
featNew user-facing or API-facing functionality
fix🐛Corrects incorrect behavior
docs📝Documentation-only changes
style💄Formatting, whitespace, semicolons
refactor♻️Restructures without changing behavior
perf⚡️Measurable performance improvement
testTest-only changes
chore🔧Build, tooling, dependencies, config
ci🚀CI/CD pipeline changes
revert⏪️Reverts a previous commit

Critical overrides (always use instead of the base emoji):

  • 💥 for ANY breaking change → 💥 feat:...
  • 🚑️ for production-critical hotfixes → 🚑️ fix:...
  • 🔒️ for security vulnerability fixes → 🔒️ fix:...

For other specialized sub-types, read references/emoji-mapping.md for 50+ emoji.

Add a body (blank line after subject) when the "why" isn't obvious, breaking changes need explanation, or the commit closes an issue (Closes #NNN).

Use direct git commit flags:

# Subject only
git commit -m "<emoji> <type>: <description>"

# Subject + body
git commit -m "<emoji> <type>: <description>" -m "<optional body>"

# Multiple body paragraphs
git commit -m "<emoji> <type>: <description>" -m "<paragraph 1>" -m "<paragraph 2>"

Step 5: Verify

  1. git log --oneline -3 — confirm the commit message matches the expected format
  2. If groups remain from Step 3, return to Step 3
  3. After all commits: git status to confirm no unintended changes remain

If commit failed (pre-commit hook, etc.): inform the user of the error and STOP. Do NOT use --amend — the failed commit was never created. On re-invocation, create a NEW commit.

Edge Cases

  • Binary files: Note in commit context; do not analyze diff content
  • Large changesets (>500 lines): Always propose splitting, even if changes appear to be a single logical unit
  • User-described intent: If the user describes the purpose of changes in the conversation, use that as primary guide but validate against the actual diff — adjust if it doesn't match

Examples

Single commit:

✨ feat: add user authentication with JWT tokens

With body:

💥 feat: migrate API response format to v2

The old format nested data under `response.data`. The new format puts it
at the top level. This is a breaking change for all API consumers.

Closes #234

Split commits (refactored shared module + added feature with tests + unrelated docs):

  1. ♻️ refactor: extract validation logic into shared module
  2. ✨ feat: add email format validation with unit tests
  3. 📝 docs: update project setup guide

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

39.99%
按下载量换算53

Claude

28.72%
按下载量换算38

Cursor

18.76%
按下载量换算25

Gemini CLI

8.92%
按下载量换算12

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills