一句话定位:Claude Code skill: four-phase plan hardening (recon, interrogate, Codex adversarial review, cross-model build & inspection) — two AI models harden your plan before a line of code exists, then swap jobs to build it. Whoever built it never grades it.。这是 GitHub 上同类项目里热度前列的一个(3.4k Star),本文按本站项目库口径做拆解:它解决什么问题、怎么装、优缺点与适合人群。口碑口径:本站未逐一亲测本文全部项目,数据来自 GitHub 公开页面(采集于 2026 年 11 月),结论以「数据事实 + 通用理性分析」为准。

它解决什么问题
AI 代码评审赛道这两年爆发,claudex-loop 的切入点是:Claude Code skill: four-phase plan hardening (recon, interrogate, Codex adversarial review, cross-model build & inspection) — two AI models harden your plan before a line of code exists, then swap jobs to build it. Whoever built it never grades it.。标签:AI。主要语言:Python,许可证:NOASSERTION,最近更新:2026-10-14——更新活跃度是判断这类项目能不能跟的第一指标,它在这一点上达标。
功能结构(据项目文档目录)
- Skills for working across Claude Code and Codex.
- Claudex Loop
- Claudex Route: standalone task routing
- The Claudex Loop workflow
- Install
- Claude Code plugin
热度与生态数据
- Star:3,417;Fork:363(Fork 数高说明二次开发的人多,生态通常更耐用)
- 仓库:github.com/chaseai-yt/claudex-loop
- 本站镜像:点击下载本站镜像(源码包)(国内直连高速,不用去 GitHub 慢慢拉) · GitHub 原仓库;镜像文件由发布流水线同步生成,若个别大仓库暂缺,点原仓库下载
- 本站拉取(Git 克隆):
git clone https://dl.yuanmazhan.net/git/claudex-loop.git——国内直连 GitHub 拉不动时用这个地址,代码与原仓库一致;本站镜像为定期同步快照,要追最新版用上方 GitHub 原仓库
安装上手
按项目官方 README 的方式起步(原文命令):
Use `/claudex-loop:claudex-route` for a lightweight recommendation or one-off handoff, or `/claudex-loop:claudex-loop`, `/claudex-loop:codex-review`, or `/claudex-loop:codex-build` for the existing workflows. Enable marketplace auto-update in the plugin menu if desired. ### Codex or manual skill installation Clone this repository and copy all skill directories together. The compatibility commands share the runtime inside `claudex-loop`; copying an alias alone is insufficient. `claudex-route` can also be installed on its own. To install **only Claudex Route**, copy `skills/claudex-route/` in
装完先把示例跑通,再动配置——这类项目九成的坑出在环境变量与模型 Key 没填对。
优点与缺点(通用理性分析)
- 优点:开源可控、社区规模摆在数据里;Python 技术栈对国内开发者友好;许可证 NOASSERTION 决定了商用边界,接入前先确认
- 缺点:这类项目迭代快、文档跟不上代码是通病;真要用于生产,模型费用、并发与安全(Key 管理、权限)都得自己兜底
适合谁
想把 AI 代码评审 落地到自己业务/社群/工作流里的开发者与站长。先收藏本页,本站会跟踪它的版本变化与插件生态,有重大更新会在文末追加。













Please log in to view comments