project-initialization
Skill2 starsChanged 2 months agoMITView on GitHub
新项目初始化选型:按项目类型选定既定脚手架基线并完成初始化。当开始新项目、选起始模板或脚手架, 或被问"用什么模板起步 / 怎么建项目"时使用——Vue3 管理后端参照 RuoYi-Vue3、Vite 轻量 Web 用 vitesse-lite、uni-app 小程序用 unibest、Nuxt / Next / React 以官方推荐初始化;产出跑通的项目骨架、 已 sync 的对应 profile 与按项目事实填写的 CLAUDE.md。
1 other project uses a version of this.
Adopted by
Most recently taken up by
- pavel-molyanov/molyanov-ai-devtheir own version, first seen Oct 25, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- pavel-molyanov/molyanov-ai-dev★ 299same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository5 other files it gives its agents.
- code-reviewSkill
- task-planningSkill
- verificationSkill
- feature-developmentSkill
- {{skill-name}}Skill
The file
Files
Project Initialization(新项目初始化选型)
基线表登记在 references/starter-baselines.md(含每个基线的布局核实与对齐点); 新基线先登记再使用,禁止临场随手选脚手架。
Purpose
把「新项目怎么起」收敛为既定基线:命中基线表的项目类型直接按基线初始化,无个人基线的栈以官方推荐为准。 产出可运行的项目骨架 + 已安装的能力配置,不产出自由发挥的技术选型。
When to Use
- 当开始一个新项目、需要选定起始模板 / 脚手架时
- 当被问「这个项目用什么模板 / 怎么初始化」时
- 当初始化后需要接入 agent-skills(sync profile、按模板填 CLAUDE.md)时
When NOT to Use
- 已有代码库的功能开发 / 改造:改用
feature-development或task-planning,不重新初始化 - 项目类型与基线都已确定的纯命令执行:直接执行,不展开选型流程
- 初始化完成后的跨端差异开发:改用
uniapp-project-patterns
Inputs
- 项目类型与形态(管理后端 / 面向用户的 Web / 小程序 / SSR 系统):缺失时先问清,禁止按猜测选型
- 发布端清单(小程序选型时):微信小程序 / H5 / App,与
uniapp-project-patterns的 Inputs 同款要求 - agent-skills 工具链可用性(
agent-skills sync):不可用时跳过 sync 步骤并在总结注明
Workflow
1. Inspect
- 对照基线表定位项目类型;读对应基线小节的布局与对齐点
- 类型未命中时进入官方推荐分支(Nuxt / Next / React 单页)
2. Plan
- 定基线:命中即用;偏离基线的选型(UI / 状态 / 请求库)逐条写理由
- 定 profile:vue →
frontend/vue、react →frontend/react、uniapp →frontend/uniapp、 nuxt →framework/nuxt、next →framework/next - 列裁剪清单(基线为全量系统时):保留 / 删除 / 后补三栏
3. Implement
- 模板类基线(vitesse-lite / unibest):以基线仓库初始化(degit / clone / 其 CLI),整套约定原样保留 (文件路由、auto-import、UnoCSS、配置化 pages / manifest)
- 参照类基线(RuoYi-Vue3):对齐其结构与模式(layout + views 分域、动态路由 + 按钮级权限、request 封装、 通用组件体系),按裁剪清单取用,不整站照搬;其以 JS 为主,TS 化是显式决策并写入项目 rules
- 官方推荐分支:Nuxt 用
npm create nuxt@latest,状态 Pinia、UI Nuxt UI;Next 用create-next-app(App Router + TypeScript);React 单页用 Vite 官方react-ts模板;官方推荐项以当前文档为准, 版本与命令联网核对,不凭记忆写 - 接入基线配置:复制 agent-skills 对应模板(
templates/<栈>/)的 4 个文件,按项目事实填写 Stack / Architecture / Commands;跑agent-skills sync --project安装能力
4. Verify
- dev 起服务打开首页成功;
build通过一次 - sync 输出的安装清单与 Plan 中定的 profile 一致(或注明工具不可用)
- CLAUDE.md 五段无模板占位残留;Commands 与
package.jsonscripts 逐条对得上
Rules
- 基线优先:命中基线表的项目类型必须用对应基线,不另起炉灶;新基线先登记进 references 再使用
- 官方优先:无个人基线的栈以官方文档当前推荐为准;禁止凭记忆编造版本号与命令
- 基线对齐但不盲抄:全量系统基线按裁剪清单取用,删除的模块同步清路由与菜单配置,不留死代码
- 初始化完成三件套缺一不可:服务跑通、profile 已 sync、CLAUDE.md 已按项目事实填写
Common Failure Modes
| 症状 | 纠正 |
跳过基线表直接 npm create 随手选型 | 回到 Inspect 对照基线表;类型未命中先补登记再动手 |
| 把 RuoYi 整站抄进小项目,带一堆用不上的模块 | 按 Plan 裁剪清单取用,删除模块同步清路由与菜单 |
| 初始化后 CLAUDE.md 还是模板占位注释 | 按 Implement 接入步骤逐段填写或删除占位 |
| 忘记 sync,能力配置缺失 | Verify 核对安装清单;工具不可用则在总结显式注明 |
| unibest 的 pages.config.ts 之外手改 pages.json | 配置化路由以 pages.config.ts 为真源,产物不手改 |
Common Rationalizations
| Agent says | Correct response |
| "官方脚手架默认项就够了" | 对照基线表与官方当前推荐逐项核对再定 |
| "CLAUDE.md 以后再填" | 初始化收尾就是填写时机,占位残留即未完成 |
| "先跑起来,profile 以后再 sync" | sync 是初始化三件套之一,不后补 |
| "基线太重,我自己搭个轻量的" | 轻量需求选基线表的轻量分支(vitesse-lite),不是绕开基线 |
Verification
- dev 服务日志 / 首页截图 + 一次
build通过的输出关键行 agent-skills sync --project的安装清单(或工具不可用的注明)- CLAUDE.md 与
package.jsonscripts 的对照结果
Exit Criteria
- 项目按基线表选定的基线(或官方推荐)初始化,偏离项附理由
- 全量基线的裁剪已执行,无死代码 / 死配置残留
- dev 起服务跑通、build 通过,证据已附
- 对应 profile 已 sync(或注明不可用),CLAUDE.md 按项目事实填写完毕
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

