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

code-review-and-commit代码审查和提交

Agent Skill

code-review-and-commit 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。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

306

周安装

13

GitHub Stars

公开资料未说明

下载量

107
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:code-review-and-commit(代码审查和提交)
来源仓库:https://github.com/abpai/skills
仓库路径:skills/code-review-and-commit
安装命令:
npx skills add https://github.com/abpai/skills --skill code-review-and-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/abpai/skills --skill code-review-and-commit

简介

code-review-and-commit 对当前变更进行高信号审查,修复关键问题并生成清晰可追溯的提交历史。

  • 适用于工作区修改检查与提交前质量保障,自动映射变更范围并应用必要修正。
  • 先 inspect 当前差异,再 fix 问题,最后 build commit plan 并请求批准后才执行推送。
  • 使用前应确认 Git 状态可用且有推送权限,避免在共享分支直接操作造成冲突。
  • 适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

SKILL.md

Code Review and Commit

Perform a high-signal review of working-tree changes, fix meaningful issues, and produce an understandable commit history.

Source basis: adapted from a local Claude Code agent prompt (code-review-and-commit.md).

Workflow

  1. Inspect current change scope.
  2. Review for correctness and maintainability issues.
  3. Apply necessary fixes.
  4. Validate updated changes.
  5. Build a commit plan.
  6. Ask for approval before creating commits.
  7. Execute commits in order and report results.

1) Inspect Current Change Scope

Run:

  • git status --short
  • git diff (unstaged)
  • git diff --staged (if relevant)

Map changes by concern (feature, fix, refactor, tests, docs) before suggesting commit boundaries.

2) Review Priorities

Prioritize in this order:

  1. Correctness and regressions.
  2. Security and secret leakage risks.
  3. Broken architecture or project-pattern violations.
  4. Missing tests for behavior changes.
  5. Readability and maintainability improvements.

Review for:

  • Logic bugs and unhandled edge cases.
  • Missing error handling or validation.
  • Performance pitfalls in changed code paths.
  • Type accuracy and docstring quality; favor concise, useful docs.
  • Low-value comments that restate obvious code behavior.
  • Resource-lifecycle problems (cleanup, context management).
  • Violations of repository conventions from project docs.

3) Apply Fixes

When findings are actionable and safe, implement fixes directly.

  • Keep scope tight to the requested work.
  • Avoid unrelated refactors unless necessary for correctness.
  • Re-check diffs after each meaningful fix.

4) Validate

Run relevant quality checks when available (for example lint, tests, type checks).

If checks cannot run, explicitly state what was skipped and why.

5) Build Commit Plan

Group changes into atomic commits that can be reverted independently.

For each proposed commit include:

  • Commit type and summary (feat, fix, refactor, test, docs, chore).
  • Exact files to stage.
  • Why this grouping is coherent.
  • Final commit message draft.

Commit message rules:

  • Use imperative mood.
  • Keep subject concise (target <= 50 chars).
  • Add body only when needed, explaining why.
  • Wrap body lines near 72 chars.

6) Approval Gate

Before running git add or git commit, present the full commit plan and request approval.

If the user asks for changes, revise the plan and re-present before executing.

7) Execute and Report

After approval:

  1. Stage only planned files for the current commit.
  2. Create the commit.
  3. Confirm success with commit hash and summary.
  4. Repeat for remaining commits.

End with a concise recap:

  • Commits created (hash + subject).
  • Files included per commit.
  • Any remaining unstaged/uncommitted changes.

Output Format

Use this structure:

  1. Review Findings grouped by severity (Critical, Important, Suggestion).
  2. Applied Fixes with file-level summary.
  3. Validation Results (commands run and outcomes).
  4. Proposed Commit Plan (numbered commits with file list + message).
  5. Execution Results after approval.

Decision Rules

  • Prefer correctness over style.
  • Favor project conventions over personal preference.
  • Surface trade-offs when multiple valid approaches exist.
  • Escalate explicitly when changes are risky or architecture-affecting.

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

34.13%
按下载量换算37

Claude

29.32%
按下载量换算31

Cursor

19.15%
按下载量换算20

Gemini CLI

8.28%
按下载量换算9

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