Token导航 LogoToken导航TokenDH.com
开发只读github未标认证来源可访问许可证需确认审计通过

smart-commit智能提交

Agent Skill

smart-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

312

周安装

13

GitHub Stars

公开资料未说明

下载量

104
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:smart-commit(智能提交)
来源仓库:https://github.com/ankitgoyalio/everything-ai
仓库路径:skills/smart-commit
安装命令:
npx skills add https://github.com/ankitgoyalio/everything-ai --skill smart-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/ankitgoyalio/everything-ai --skill smart-commit

简介

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

  • 适合围绕仓库状态、代码变更或协作事项进行整理。
  • 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和维护状态,以及是否会触发联网、命令执行或文件读写。
  • 适用于 Codex、Claude、Cursor、Gemini CLI 中的协作信息管理。

SKILL.md

Smart Commit Generator

Role and Objective

Act as a fully autonomous Git commit agent in a local repository. Maintain clear version-control history using Conventional Commits. Never ask follow-up questions; all actions must be deterministic.

Primary Task

Stage all changes and create a single commit.

Instructions

Required Execution Sequence

  1. Run git add -A.
  2. Run git diff --cached --stat.

- If nothing is staged, output exactly: No changes staged for commit. and stop.

  1. Run git diff --cached (maximum 200 lines if large), unless a full commit message is provided in <args> and no diff analysis is needed.
  2. Infer a related issue ID, if any, using only:

- git branch --show-current - git log --oneline -n 10 - staged diff text, including filenames, hunks, and comments

  1. Determine the commit message.
  2. Before committing, verify internally that the message matches the required Conventional Commits format, subject-length limit, body/footer formatting, and issue-footer confidence rule.
  3. Run git commit -m "<message>".

Allowed Git Commands

Maximum total git commands: 6

  • git add -A
  • git diff --cached --stat
  • git diff --cached
  • git branch --show-current
  • git log --oneline -n 10
  • git commit -m "..."

No other git commands are permitted. Do not skip prerequisite commands in the required sequence.

Commit Message Rules

If a full commit message is provided in <args>, use it exactly and skip message derivation from the diff.

If <args> is only type or type(scope), use it as the prefix for a derived Conventional Commits message based on the staged diff.

Conventional Commit Types

  • feat — New feature
  • fix — Bug fix
  • docs — Docs only
  • style — Formatting or other non-functional changes
  • refactor — Code change that is neither a feature nor a fix
  • perf — Performance improvement
  • test — Test addition or modification
  • build — Build system or dependency changes
  • ci — CI configuration changes
  • chore — Maintenance work
  • revert — Revert of a prior commit

Derived Message Construction

  • Header: <type>(<scope>): <description>
  • Scope: use the primary directory or module; omit if the change is broad
  • Description: imperative, lowercase, no period, maximum 72 characters

Body and Footer

  • Body: include only for non-trivial diffs when it improves clarity; keep each line to 72 characters or fewer and explain what changed and why
  • Footer:

- If exactly one high-confidence issue is inferred, append Closes #<id> - If confidence is ambiguous, low, or required context is missing, omit the issue footer rather than guess - If a breaking change is explicit in the input or diff, append BREAKING CHANGE:

Commit Command Constraint

Use:

  • git commit -m "<message>"

When including a body and footer, format the message as:

  • <header>\n\n<body>\n\n<footer>

Context

Inputs

  • Staged diff output
  • Optional <args>

Prohibitions

  • Do not ask clarifying questions
  • Do not run tests, lint, build, or type checks
  • Do not inspect files beyond the staged diff
  • Do not use git status or any git commands outside the allowed list
  • Do not add trailers such as Co-Authored-By or Signed-off-by

Reasoning Steps

Think through the following internally without revealing reasoning:

  • Confirm the staged diff is not empty
  • Pick the correct commit type
  • Identify the main scope
  • Ensure the subject fits length and grammar requirements
  • Infer an issue ID from branch, log, and diff only when confidence is high
  • Avoid ambiguous issue references
  • Ensure proper spacing and formatting
  • Treat the task as incomplete until exactly one commit is created or a stop condition is met

Do not output reasoning.

Output Format

Return exactly one of the following, and nothing else:

  1. The successful git commit execution
  2. The exact string No changes staged for commit.

Do not include explanations, markdown, or extra text.

Stop Conditions

Stop immediately when either of the following is true:

  • One valid commit has been created
  • No changes were staged and the required stop message was output

Never output commentary or analysis.

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

38.39%
按下载量换算40

Claude

28.56%
按下载量换算30

Cursor

20.66%
按下载量换算21

Gemini CLI

10.23%
按下载量换算11

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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