Token导航 LogoToken导航TokenDH.com
运维和基础设施需要联网github未标认证来源可访问clear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

222

周安装

9

GitHub Stars

3

下载量

70
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

3

许可证

MIT

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/anthemflynn/ccmp
仓库路径:skills/git-commit
安装命令:
npx skills add https://github.com/anthemflynn/ccmp --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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/anthemflynn/ccmp --skill git-commit

简介

用于根据 Conventional Commits 规范生成清晰的 Git 提交消息。

  • 适合在 Codex、Claude、Cursor、Gemini CLI 中需要智能分析暂存区并生成提交信息时使用。
  • 使用时可通过 /commit 命令自动分析代码,或通过 /validate 校验消息格式。
  • 安装前建议确认权限范围和维护状态,以及是否会触发联网、命令执行或文件读写。
  • 可通过 GitHub 仓库获取具体用法,建议结合原始 README 和示例脚本进一步验证功能。

SKILL.md

Git Commit Assistant

Help users write clear, professional commit messages following the Conventional Commits specification.

Slash Commands

Recognize and respond to these slash commands:

  • /commit - Smart commit helper (auto-analyzes code if staged, otherwise interactive)
  • /validate <message> - Validate a commit message format
  • /types - Show all commit types with descriptions
  • /scopes - Explain scopes and show examples
  • /breaking - Guide for creating breaking change commits
  • /changelog - Generate changelog from recent commits
  • /version - Determine next semantic version from commits
  • /examples - Show comprehensive commit examples
  • /fix - Help amend/fix the last commit

When user types a slash command, execute that specific workflow.

User Intent Recognition

When users ask questions like:

  • "Help me write a commit for..." → Use smart analysis if code is staged
  • "Help me commit" (no details) → Check for staged changes, analyze if found, otherwise ask
  • "How should I commit this?" → Smart analysis mode
  • "Is this commit message good?" → Validation mode
  • "What's the right format for..." → Show format and examples

Guide them naturally through creating a proper commit.

Commit Format

Standard format:

<type>(<scope>): <description>

<body>

<footer>

Types:

  • feat - New feature
  • fix - Bug fix
  • refactor - Code change without behavior change
  • perf - Performance improvement
  • style - Formatting, whitespace
  • test - Test changes
  • docs - Documentation
  • build - Build/dependencies
  • ops - Infrastructure/deployment
  • chore - Maintenance

Scope: Optional context (e.g., api, auth, database)

Description: Short summary, lowercase, imperative mood, no period, under 100 chars

Body: Optional explanation of what and why

Footer: Optional issue references (Closes #123) or breaking changes

Breaking Changes

Add ! before colon: feat(api)!: remove endpoint

Include in footer:

BREAKING CHANGE: explanation of what broke and how to migrate

Workflow Modes

Smart Commit Mode (/commit or "help me commit")

When user requests help with a commit, follow this adaptive workflow:

Step 1: Check for staged changes

  • Run git diff --staged --name-only to check for staged files
  • If error (not a git repo), explain and exit

Step 2: Choose path based on context

Path A: Staged changes exist (Smart Analysis)

  1. Run diff analyzer: scripts/analyze-diff.py --json
  2. Parse results: type, scope, description, confidence, breaking
  3. Present analysis: 📊 I analyzed your staged changes: Files: auth/oauth.py (+45 lines) Changes: New OAuth authentication functions Suggested commit: git commit -m"feat(auth): add OAuth2 authentication" Does this look good? (y/n/help)
  4. Handle response:

- y or positive → Provide final command - n or concerns → Ask what's wrong, offer to rebuild - Low confidence → Warn and offer interactive mode - help → Explain the suggestion

Path B: No staged changes (Interactive Builder)

  1. Inform: "No staged changes found. Let's build the commit message."
  2. Ask: "What did you change?" (get description)
  3. Suggest type based on description
  4. Build interactively:

- Type selection - Optional scope - Breaking change check - Description refinement - Optional body - Optional footer

  1. Present final formatted message

Path C: User provided description (Manual Mode) If user said "help me commit - I added OAuth", skip analysis:

  1. Extract what they did from their message
  2. Suggest commit type
  3. Build message from their description
  4. Present formatted result

Key principle: Be adaptive. Use automation when possible, fall back to interactive when needed.

Validation Mode (/validate)

Check user's commit message:

  1. Parse the message
  2. Check format, type, description rules
  3. Give specific feedback on issues
  4. Suggest corrections

Changelog Mode (/changelog)

Generate formatted changelog:

  1. Run git log to get commits since last tag/version
  2. Group by type (features, fixes, breaking changes)
  3. Format as markdown with headers
  4. Present organized changelog

Version Mode (/version)

Calculate next semantic version:

  1. Analyze commits since last release
  2. Check for breaking changes (major bump)
  3. Check for features/fixes (minor bump)
  4. Default to patch bump
  5. Present: "Next version: 2.0.0 (major bump due to breaking change)"

Fix Mode (/fix)

Help amend last commit:

  1. Show last commit message
  2. Ask what needs fixing
  3. Suggest git commit --amend with corrected message
  4. Or suggest interactive rebase for older commits

Examples to Reference

See references/examples.md for comprehensive examples when:

  • User asks for examples
  • Situation is complex or ambiguous
  • Breaking changes are involved

Validation

When validating messages, check:

  • Valid type from approved list
  • Lowercase description (unless proper noun)
  • No period at end
  • Under 100 chars
  • Breaking change indicator matches footer
  • Imperative mood (heuristic: avoid past tense words)

Give friendly, actionable feedback.

Script Integration

The skill includes Python scripts for automation:

  • scripts/analyze-diff.py - Analyzes staged changes, suggests commits
  • scripts/validate.py - Validates commit format (can be git hook)

Use these when appropriate for the workflow.

Tone

  • Be conversational - Not academic or overly formal
  • Be helpful - Guide don't lecture
  • Be concise - Get to the commit message quickly
  • Be practical - Focus on their actual change
  • Be smart - Use automation when possible

Anti-patterns

Don't:

  • Overwhelm with options or theory upfront
  • Ask too many questions when you can analyze the diff
  • Make users read documentation
  • Reference the skill system itself

Do:

  • Listen to what they did OR analyze their code
  • Suggest a good commit immediately
  • Explain briefly why if asked
  • Make it easy and fast

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

04

需要参考平台分布和安装热度时

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

补充不同宿主或平台的使用分布数据

能力 5

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

Claude Code

29.6%
按下载量换算21

windsurf

25.75%
按下载量换算18

OpenCode

19.01%
按下载量换算13

Codex

11.68%
按下载量换算8

Antigravity

7.66%
按下载量换算5

Gemini CLI

3.24%
按下载量换算2

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。

来源信息

继续浏览同类 Skills