‹ 返回总目录
开发工具 最近更新08-05

coding-principles

| 2026-08-01 收录

☆ 收藏
🛒 📚 AI提效实战指南 · 图解Skill🛒 📱 大流量手机卡 · 低月租📣 加入推广 · 佣金80-300/张 ›

🛍️ 更多精选好物 ›

📊 公开数据一览

📦 榜单安装量66
⭐ GitHub Stars664(fork 101)
🗓️ 最近推送2026-08-05
🌱 项目创建2025-10-10
💻 语言 / LicenseJavaScript · MIT
🐛 开放Issue0

📝 工具简介

由专业AI agents驱动的生产级开发工作流,专为Claude Code设计。

📖 怎么用

先启动 Claude Code,输入 `claude` 进入交互界面。接着添加插件市场:`/plugin marketplace add shinpr/claude-code-workflows`。按需安装工作流:后端或通用场景跑 `/plugin install dev-workflows@claude-code-workflows`,前端用 `dev-workflows-frontend`,全栈则用 `claude plugin install dev-workflows-fullstack@claude-code-workflows --scope project`。装完执行 `/reload-plugins` 重载,然后直接调 `/recipe-implement "Add rate limiting to the public API"` 这类命令开始干活。流程会先检查现有代码、设计最小改动、暂停等你批准,再逐个任务实现,最后核对是否跑偏。

📋 迷你测评

面向Claude Code的生产级开发工作流,由专业AI agents驱动。亮点是强调“生产级”,适合正经项目。没提具体功能细节,定位偏流程规范。

📄 README 要点

这是一套给 Claude Code 用的可重复软件开发工作流,核心是让设计决策在实现、测试和审查全过程中都可追踪。每个阶段都在全新的 agent 上下文中运行,通过明确的产物进行交接。工作流会先检查现有代码库,设计最小且足够的改动,暂停等待批准,然后一次只实现一个任务,最后检查完成的工作是否仍然符合约定的范围和需求。对于跨文件、跨层或跨贡献者的生产级改动,建议使用端到端配方,因为长时间的 AI 编码会话容易悄悄扩大范围或在设计与实现之间丢失重要决策。还有聚焦配方,可以让你在设计后停止、从已批准的产物继续、审查实现或在不改动代码的情况下调查问题。简单说,就是给 AI 编程加个流程管控,防止跑偏。

查看英文原文
Claude Code Development Workflows Repeatable software development workflows for Claude Code that keep design decisions traceable through implementation, tests, and review. Each phase runs in a fresh agent context and hands off through explicit artifacts. The workflow inspects the existing codebase, designs the smallest sufficient change, pauses for approval, implements one task at a time, and checks whether the finished work still matches the agreed scope and requirements. Use the end to end recipes for production changes that span files, layers, or contributors, where a long AI coding session can quietly grow in scope or lose an important decision between design and implementation. Focused recipes let you stop after design, continue from approved artifacts, review an implementation, or investigate a problem without changing the code. Why use a development workflow? Claude Code can complete an individual coding task well. The harder problem is keeping a larger change coherent after the first answer. Suppose analysis finds that an existing authentication path should be extended. Halfway through implementation, a new abstraction looks convenient, the response contract changes with it, and the frontend adapts to the new shape. Every local edit may look reasonable and every test may pass, while the result no longer matches the design that the team agreed on. In that example, extending the existing authentication path remains the default. A second mechanism needs evidence during design, and the agreed response contract is carried into the work plan, implementation tasks, tests, and final review. If implementation discovers that the contract really must change, the workflow stops and returns to design instead of letting the frontend quietly adapt to a decision nobody reviewed. Because the process is packaged as a Claude Code plugin, a team can apply it consistently across repositories. Quick Start Requires a Claude Code release with plugin marketplace support. Choose a pa

💬 评论(0)

交流使用体验、避坑建议;违规内容将被删除

💬 意见反馈 / 联系客服

数据来源:skills.sh 榜单 + GitHub 公开数据,非人工实测,仅供参考