Token导航 LogoToken导航

Agent Skill · 研究检索

review-changes(审查变更)

review-changes 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。

review-changes 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为需要联网;许可证信息还不明确,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 38,198。

总安装 1,597周安装 64GitHub Stars 公开资料未说明下载 517核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

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:review-changes(审查变更)
来源仓库:https://github.com/tahajemmali/skills
仓库路径:skills/review-changes
安装命令:
npx skills add https://github.com/tahajemmali/skills --skill review-changes
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/tahajemmali/skills --skill review-changes

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

review-changes 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。适用于基于关键词或任务场景审查变更内容的信息提取。通过 npx skills add 命令从指定 GitHub 仓库安装并使用。

建议确认权限范围和维护状态,避免触发联网或文件读写操作。可结合原始 README 进一步核验具体功能和调用方式。

来源仓库内路径:skills/review-changes

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/tahajemmali/skills --skill review-changes;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要需要联网,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

review-changes 怎么安装?

在终端执行 npx skills add https://github.com/tahajemmali/skills --skill review-changes。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

review-changes 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

review-changes 需要什么权限?

权限级别是需要联网。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。

review-changes 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

review-changes 用的人多吗?

公开数据是累计安装 1,597 次,近一周安装 64 次,下载 517 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

review-changes 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
review-changes

审查变更 · 本页

1,597—需要联网未明确通过

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Review Changes

Overview

Review recent code changes against the repo's documented guidelines.

Default to review mode: identify concrete violations, risks, regressions, and missing tests first. Only give a brief clean bill of health when no meaningful findings exist.

Workflow

  1. Inspect the latest change set.
  2. Read the relevant parts of the project's guidelines (AGENTS.md, CLAUDE.md, or any onboarding doc referenced there).
  3. Read the changed code directly.
  4. Compare the implementation against repo rules, not generic style opinions.
  5. Report findings first, ordered by severity, with file references.
  6. If no findings exist, say that explicitly and note any residual risk or verification gap.

Inspect The Change Set

Prefer the smallest source of truth that answers the review request:

  • git status --short
  • git diff --stat
  • git diff -- <path>
  • git log -1 --stat
  • git show <commit>

If the user asks about committed work, review the relevant commit or commit range instead of the working tree.

Do not review from summaries alone. Read the affected files.

Review Focus

Use the project's documented guidelines as the review standard. Focus especially on:

  • correct file and feature placement per the project's directory conventions
  • DRY and reuse of logic, components, and flows
  • state management following the project's chosen patterns
  • dependency injection and service resolution following project conventions
  • repository and data access contracts matching established patterns
  • tests added or updated when behavior changes
  • files growing beyond roughly 800 lines without being split

Do not invent standards beyond what the repo documents unless a change is obviously broken on its own terms.

Repo-Specific Checks

Treat these as high-value checks:

  • duplicated components or near-identical UI that should be shared or parameterized
  • new services or repositories instantiated ad hoc inside views/components
  • raw exceptions escaping into state management or UI layers
  • hardcoded user-facing text that should be localized (if the project uses i18n)
  • behavior changes without matching test updates
  • large files that should be split into focused files

Use the checklist below when you need to scan quickly.

Review Checklist

Architecture

  • Does the code belong in the shared/core layer or in the feature that owns it?
  • Did the change duplicate a shared concern instead of reusing existing code?
  • Did any file grow beyond roughly 800 lines without being split?

State Management

  • Does the state management approach follow the project's chosen pattern?
  • If a new pattern was introduced, is there real complexity to justify it?

Data And DI

  • Are dependencies resolved via the project's DI system instead of ad hoc construction?
  • Do repositories and data access layers follow the established contracts?
  • Do mock or test implementations still match the real contract?

UI And Reuse

  • Is there an existing component that should have been reused?
  • Could the new UI be parameterized instead of cloned?
  • Should the component live in shared code because multiple features can use it?

User-Facing Behavior

  • Is all visible text localized (if the project uses i18n)?
  • Were any persistence changes made in a way that could create partial or inconsistent state?

Tests

  • Did behavior change without test updates?
  • If a reusable abstraction changed, were existing callers protected with tests?
  • Is there any residual risk because validation was not run?

Output Format

Findings are the primary output.

For each finding:

  • state the severity
  • state the problem
  • explain why it violates the project's guidelines or creates a concrete risk
  • include a file reference
  • suggest the fix direction briefly

After findings, optionally include:

  • open questions or assumptions
  • a short summary if needed

If there are no findings, say that explicitly. Mention residual risks such as unrun tests, incomplete diff scope, or unverified runtime behavior.

Review Discipline

  • Prefer repo-specific reasoning over generic best-practice language.
  • Do not request documentation updates unless the user asks for doc maintenance separately.
  • Do not suggest broad rewrites when a focused fix would satisfy the repo rule.
  • Be strict about architectural drift, duplication, and missing tests.
  • Be careful with false positives: confirm the code actually violates the rule before flagging it.

平台分布

Codex

33.86%
按下载量换算175

Claude

27.97%
按下载量换算145

Cursor

18.59%
按下载量换算96

Gemini CLI

9.77%
按下载量换算51

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。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills