Token导航 LogoToken导航TokenDH.com
前端设计敏感数据github未标认证来源可访问许可证需确认审计通过

git-commitsgit 提交

Agent Skill

git-commits 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

870

周安装

37

GitHub Stars

公开资料未说明

下载量

305
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-commits(git 提交)
来源仓库:https://github.com/loxosceles/ai-dev
仓库路径:skills/git-commits
安装命令:
npx skills add https://github.com/loxosceles/ai-dev --skill git-commits
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/loxosceles/ai-dev --skill git-commits

简介

git-commits 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 中整理协作事项。

  • 适用于 Git 提交记录分析、版本控制和代码协作流程管理等前端设计场景。
  • 围绕仓库状态和代码变更进行信息整合,需结合原始 README 确认具体用法。
  • 安装命令为 npx skills add https://github.com/loxosceles/ai-dev --skill git-commits。
  • 使用前建议确认权限范围、维护状态及是否触发联网或文件读写操作。

SKILL.md

Git Commits

This is a strict guideline. Follow these rules exactly.

How to format and create git commits, branches, and PRs when helping developers.


🚨 CRITICAL: Security Check FIRST

Before analyzing ANY commits, check for sensitive files:

NEVER commit these files:

Environment & Configuration:

  • .env (any variant: .env, .env.local, .env.prod, .env.dev, .env.test, .env.*)
  • .envrc, .env.example (if contains real values)
  • config.json, secrets.json, credentials.json
  • .npmrc, .pypirc (if contains auth tokens)

AWS & Cloud Credentials:

  • credentials, config (in .aws/ directory)
  • serviceAccount.json, gcloud-credentials.json
  • terraform.tfvars (if contains secrets)
  • pulumi.*.yaml (if contains secrets)

Private Keys & Certificates:

  • *.pem, *.key, *.p12, *.pfx
  • id_rsa, id_ed25519, *.pub (private keys only)
  • *.crt (if paired with private key)
  • keystore.jks, truststore.jks

Database & API:

  • Files with connection strings containing passwords
  • API keys, access tokens, OAuth secrets
  • Database dumps with real data
  • *.sql files with production data

Other Sensitive:

  • .git-credentials
  • .netrc
  • *.log files with sensitive data
  • Backup files with credentials (*.bak, *.backup)

If detected:

  1. STOP immediately
  2. Alert developer with 🚨 WARNING
  3. List the sensitive files found
  4. Refuse to proceed until removed from staging
  5. Remind developer to add to .gitignore

Example alert:

🚨 SECURITY WARNING: Sensitive files detected!

The following files should NEVER be committed:
- infrastructure/.env.prod (environment file with credentials)
- .aws/credentials (AWS access keys)
- config/database.json (database password)

ACTION REQUIRED:
1. Remove from staging: git reset HEAD <file>
2. Add to .gitignore if not already present
3. Verify no secrets were previously committed
4. Consider rotating any exposed credentials

I cannot proceed with commit suggestions until these are resolved.

Commit Message Format

Structure:

<type>: <description>

Rules:

  • One logical change per commit
  • Single line only (no multi-line commits)
  • Imperative form: "Add", "Create", "Fix", "Update" (not "Added", "Adding")
  • Be specific but concise

Types:

  • feat - New feature
  • fix - Bug fix
  • refactor - Code restructuring (no behavior change)
  • docs - Documentation only
  • style - Formatting, whitespace
  • test - Adding/updating tests
  • chore - Build, config, dependencies

Examples:

feat: Add user authentication flow
fix: Resolve database connection timeout
docs: Update API endpoint documentation
refactor: Extract validation logic to separate function

Commit Grouping Strategy

Analysis Criteria

Group files together when they:

  • Implement the same feature
  • Fix the same bug
  • Are part of the same refactoring
  • Have strong dependencies (one requires the other)

Separate files when they:

  • Serve different purposes (feature vs refactor vs docs)
  • Touch different features/components
  • Can work independently
  • Represent different logical changes

Grouping by Purpose

Analyze changes by:

  • Purpose: What problem does this solve?
  • Scope: What area of the codebase?
  • Independence: Can this stand alone?
  • Dependencies: Does it depend on other changes?

Common patterns:

  • New feature = components + types + utilities + docs
  • Refactoring = code quality improvements separate from features
  • Bug fix = minimal files, specific to issue
  • Config = package.json, environment, build files

Process

When developer asks for commit help:

  1. 🚨 Security Check - Verify no .env files or credentials in changes (STOP if found)
  2. Analyze changes - Run git commands to see actual changes (don't guess)
  3. Group logically - Organize by purpose, not file type
  4. Generate messages - Follow chosen format (simple or conventional)
  5. Present for approval - Show commit groups with affected files
  6. Wait for approval - Developer stages and commits in their tool

Example Output

Simple Format

Commit 1: Add user authentication logic
Files:
- auth.ts (authentication functions)
- types.ts (auth types)

Commit 2: Update database schema
Files:
- schema.ts (user table changes)

Conventional Commits

📦 Commit 1: feat: Add user authentication flow
Files:
- frontend/contexts/auth-context.tsx (auth state management)
- frontend/hooks/use-auth.ts (auth hook)
- frontend/types/auth.ts (type definitions)
Group: Authentication feature - all related

📦 Commit 2: refactor: Extract API client utils
Files:
- frontend/lib/api-client.ts (extracted from inline code)
- frontend/utils/fetch-helper.ts (helper functions)
Group: Code quality - independent refactoring

📦 Commit 3: docs: Update authentication setup guide
Files:
- docs/guides/authentication.md (auth flow documentation)
Group: Documentation - separate from code

Branches and PRs

Branch naming:

  • Use descriptive, kebab-case names with type prefix: feat/user-profile, fix/auth-timeout
  • NEVER reference planning phases: No phase-1, step-2-3, etc.
  • Planning docs are not tracked — references are meaningless in git history

PR titles — plain descriptive (NOT conventional commits):

  • Describe what the PR does in plain English
  • No feat:, fix:, chore: prefixes — the branch name and labels carry the type
  • Imperative form, capital letter
# ✅ Good PR titles
Add scales browser UI
Fix test runner in CI
Migrate devcontainer to ai-standards

# ❌ Bad PR titles (don't use conventional commit format)
feat: Add scales browser UI
chore: Migrate devcontainer to ai-standards

# ❌ Bad PR titles (auto-generated GitHub defaults)
Dev
Fix/auth timeout on mobile

Why no conventional commits on PRs? Versioning is label-driven (major/minor/patch labels on PRs to main). The PR title is for humans — keep it clean.

Examples:

# ✅ Good branch names
git checkout -b feat/user-profile
git checkout -b fix/auth-timeout
git checkout -b update-api-endpoints

# ❌ Bad branch names (no planning references)
git checkout -b phase-1-setup
git checkout -b step-2-3-implement-auth
git checkout -b task-1-2-database

Versioning:

  • Handled automatically by GitHub Action on merge to main
  • PR labels (major, minor, patch) determine version bump
  • Never manually write version numbers in commits or PR titles

Guidelines

  • Security first: Always check for .env files and credentials before proceeding
  • Analyze first: Always run git status and git diff to see actual changes
  • Logical grouping: Group by feature/fix, not by file type
  • No line ranges: Just list files, not specific line numbers
  • Wait for approval: Don't execute commits without explicit approval
  • No planning references: Never mention "Step 1.2", "Phase 3", etc. in commits, branches, or PRs
  • Check for issues: Flag .env files, console.log in production, missing newlines

What NOT to Do

NEVER commit.env files or credentials - STOP and alert if detected ❌ Don't stage files (developer does this) ❌ Don't commit (developer does this) ❌ Don't push (developer controls when) ❌ Don't modify files unless asked ❌ Don't guess at changes - always analyze actual diffs


Progressive Improvement

If the developer corrects a behavior that this skill should have prevented, suggest a specific amendment to this skill to prevent the same correction in the future.

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

38.63%
按下载量换算118

Claude

28.53%
按下载量换算87

Cursor

17.49%
按下载量换算53

Gemini CLI

9.96%
按下载量换算30

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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