domain-init

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Domain Registry Initializer

域注册表初始化工具

Initialize or refresh the Domain Registry in this project's

CLAUDE.md

初始化或更新本项目

CLAUDE.md

中的Domain Registry。

What This Does

功能说明

Scans the project for library/domain directories
Reads each library's metadata (
```
project.json
```
,
```
package.json
```
, or
```
ng-package.json
```
)
Writes (or updates) a
```
## Domain Registry
```
section in
```
CLAUDE.md
```
Adds the auto-load instruction if it's not already present

扫描项目中的库/域目录
读取每个库的元数据（
```
project.json
```
、
```
package.json
```
或
```
ng-package.json
```
）
在
```
CLAUDE.md
```
中写入（或更新）
```
## Domain Registry
```
章节
如果自动加载说明不存在，则添加该说明

Instructions

使用说明

Step 1 — Detect Project Structure

步骤1 — 检测项目结构

Run the detection script to find all library directories:

bash

bash .claude/skills/domain-init/scripts/scan_domains.sh 2>/dev/null \
  || node .claude/skills/domain-init/scripts/scan_domains.js 2>/dev/null

If neither script is available, scan manually:

Look for directories under
```
libs/
```
that contain
```
project.json
```
or
```
package.json
```

For Nx workspaces: read

project.json

→

name

and

targets.build.options.project

For plain npm: read
```
package.json
```
→
```
name
```
For Angular libraries: read
```
ng-package.json
```
→ check parent
```
package.json
```
for name

运行检测脚本查找所有库目录：

bash

bash .claude/skills/domain-init/scripts/scan_domains.sh 2>/dev/null \
  || node .claude/skills/domain-init/scripts/scan_domains.js 2>/dev/null

如果两个脚本都不可用，请手动扫描：

查找
```
libs/
```
目录下包含
```
project.json
```
或
```
package.json
```
的目录

对于Nx工作区：读取

project.json

→

name

和

targets.build.options.project

字段

对于普通npm项目：读取
```
package.json
```
→
```
name
```
字段
对于Angular库：读取
```
ng-package.json
```
→ 检查上级
```
package.json
```
获取名称

Step 2 — Build the Registry Table

步骤2 — 构建注册表表格

For each found library, collect:

Field	Source
`Path`	Relative directory path from workspace root
`Import Path`	`importPath` from `project.json` , or `name` from `package.json`
`Description`	First line of existing `README.md` (skip `#` heading), or generate from library name

Generate description heuristics if no README exists:

```
widget-*
```
→ "Widget: [name] feature module"
```
*-domain
```
or
```
domain
```
→ "Domain layer: business logic and state"
```
*-feature
```
or
```
feature
```
→ "Feature layer: container components"
```
*-ui
```
or
```
ui
```
→ "UI layer: presentational components"
```
*-util
```
or
```
util
```
→ "Utilities and helpers"
```
shared/*
```
→ "Shared: [name]"

针对每个找到的库，收集以下信息：

字段	来源
`Path`	工作区根目录下的相对目录路径
`Import Path`	`project.json` 中的 `importPath` ，或 `package.json` 中的 `name`
`Description`	现有 `README.md` 的第一行（跳过 `#` 标题），或根据库名生成

如果没有README文件，可通过启发式规则生成描述：

```
widget-*
```
→ "Widget: [name] 功能模块"
```
*-domain
```
或
```
domain
```
→ "域层：业务逻辑与状态"
```
*-feature
```
或
```
feature
```
→ "功能层：容器组件"
```
*-ui
```
或
```
ui
```
→ "UI层：展示组件"
```
*-util
```
或
```
util
```
→ "工具函数与辅助方法"
```
shared/*
```
→ "共享：[name]"

Step 3 — Write to CLAUDE.md

步骤3 — 写入CLAUDE.md

Read the current

CLAUDE.md

. Then:

If

already exists: Replace everything between the start and end markers with the new table.

If it does not exist: Append the following block to

CLAUDE.md

markdown

undefined

读取当前的

CLAUDE.md

，然后按以下规则处理：

如果

已存在：替换起止标记之间的所有内容为新生成的表格。

如果不存在： 将以下代码块追加到

CLAUDE.md

末尾：

markdown

undefined

Domain Registry

Auto-maintained by domain-init skill. Run
/domain-init
to regenerate.

Path	Import Path	Description
<path>	<import-path>	<description>

undefined

Auto-maintained by domain-init skill. Run
/domain-init
to regenerate.

Path	Import Path	Description
<path>	<import-path>	<description>

undefined

Step 4 — Add Auto-Load Instruction (if missing)

步骤4 — 添加自动加载说明（如缺失）

Check if

CLAUDE.md

already contains the phrase

domain-init skill

Domain Registry

. If it does NOT, also append:

markdown

undefined

检查

CLAUDE.md

中是否已包含

domain-init skill

或

Domain Registry

字样。如果没有，同时追加以下内容：

markdown

undefined

Domain Context Loading

The

domain-init

skill is active. Before working on any library listed in the Domain Registry above, silently read its

README.md

for architecture context.

undefined

The

domain-init

skill is active. Before working on any library listed in the Domain Registry above, silently read its

README.md

for architecture context.

undefined

Step 5 — Report

步骤5 — 结果报告

After writing, output a summary:

✓ Domain Registry updated in CLAUDE.md
  Found N domains:
  - libs/widget-chat  →  @scope/widget-chat
  - libs/shared/...   →  @scope/shared/...
  ...

  Domains without README.md (consider running /domain-readme <path>):
  - libs/dashboards-manager
  - libs/widget-map

写入完成后，输出汇总信息：

✓ Domain Registry updated in CLAUDE.md
  Found N domains:
  - libs/widget-chat  →  @scope/widget-chat
  - libs/shared/...   →  @scope/shared/...
  ...

  Domains without README.md (consider running /domain-readme <path>):
  - libs/dashboards-manager
  - libs/widget-map

Arguments

参数说明

$ARGUMENTS

— optional scan root path. Defaults to

libs/

if empty.

Examples:

```
/domain-init
```
— scans
```
libs/
```
```
/domain-init libs/shared
```
— scans only
```
libs/shared/
```
```
/domain-init packages
```
— scans a
```
packages/
```
monorepo structure

$ARGUMENTS

— 可选的扫描根路径。如果为空默认是

libs/

。

示例：

```
/domain-init
```
— 扫描
```
libs/
```
目录
```
/domain-init libs/shared
```
— 仅扫描
```
libs/shared/
```
目录
```
/domain-init packages
```
— 扫描monorepo结构的
```
packages/
```
目录

Re-running

重复运行

This skill is idempotent. Running it multiple times only updates the registry table, preserving all other content in

CLAUDE.md

本skill是幂等的。多次运行只会更新注册表表格，不会修改

CLAUDE.md

中的其他内容。