domain-init

Original：🇺🇸 English

Translated

3 scripts

Initializes or refreshes the Domain Registry in CLAUDE.md by scanning the project for library directories. Run this once per project, or after adding new libraries. Creates the registry table that domain-context skill uses for auto-loading.

2installs

Sourcebogusweb/workspace-context-skills

Added on2026-03-03

NPX Install

npx skill4agent add bogusweb/workspace-context-skills domain-init

SKILL.md Content

View Translation Comparison →

Domain Registry Initializer

Initialize or refresh the Domain Registry in this project's

CLAUDE.md

.

What This Does

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

Instructions

Step 1 — Detect Project Structure

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

Step 2 — Build the Registry Table

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]"

Step 3 — Write to 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


## Domain Registry

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

<!-- domain-registry-start -->
| Path | Import Path | Description |
|------|-------------|-------------|
| <path> | <import-path> | <description> |
<!-- domain-registry-end -->

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

Check if

CLAUDE.md

already contains the phrase

domain-init skill

or

Domain Registry

. If it does NOT, also append:

markdown


## 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.

Step 5 — Report

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

Arguments

$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

Re-running

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

CLAUDE.md

.