Use when reviewing UI for accessibility — WCAG 2.2 AA, keyboard nav, focus, ARIA, contrast, screen-reader semantics — even on 'is this a11y-OK?' or 'mach das barrierefrei'.
日本語の概要は準備中です。原文の説明を表示しています。
When working within any module under `modules.root_paths` from `.agent-project-settings.yml` — Laravel HMVC, Symfony DDD-lite, Node monorepo, Python src/, Go internal/, or a custom path.
インストールする前に、エージェントに与えられる指示の中身を確認できます。
Use this skill when creating, exploring, or working within a module under
any directory listed in modules.root_paths (team setting in
.agent-project-settings.yml). The skill is stack-agnostic — Laravel
HMVC, Symfony DDD-lite, Node monorepo, Python src layout, Go internal/,
or a custom path all flow through the same procedure with a
stack-specific carve-out at the bottom.
When modules.enabled is false (the default) the skill is a no-op —
the project does not opt into module-aware behavior.
Before creating, renaming, or modifying anything inside a module:
modules: block from .agent-project-settings.yml —
never assume a specific module root.modules.root_paths entry and
review the target module's README.md (or package.json /
pyproject.toml description) to understand its purpose.{module_root}/{ModuleName}/{modules.agent_folder}/contexts/.modules.namespace_template
shape and confirm the conventions before generating files.Skip this step → risk creating files in the wrong root, breaking stack-native auto-loading, or duplicating existing modules.
modules: block from .agent-project-settings.yml via the
loader (get_modules_config() in scripts/_lib/agent_settings.ts).modules.root_paths, read the module's README.md
(or package.json / pyproject.toml description) for purpose.{module_root}/{modules.agent_folder}/ (default agents/), read them.modules.namespace_template shape
or root-path heuristic — apply stack-specific conventions on top of
the generic procedure.modules: block is empty)When modules.enabled is unset / false, the skill consults the
auto-detection table in
/module explore Step 1 — same six
stack shapes, fallback only. The skill never writes the modules:
block automatically; that is propose_modules_config.ts plus user
confirmation per
/agents init Step 7.
{module_root}/{ModuleName}/
├── <stack-native source layout> # see carve-out below
├── tests/ # module-specific tests
├── README.md # module description
└── {modules.agent_folder}/ # default: agents/
├── features/ # module-scoped feature plans
├── roadmaps/ # module-scoped roadmaps
└── contexts/ # module-scoped context docs
The agent-folder name comes from modules.agent_folder (default
agents). Skip directories listed in modules.skip_dirs (default
.module-template, .example).
Apply the section that matches the project's
modules.namespace_template and root-path layout.
Triggers when: modules.namespace_template starts with
App\Modules\ or root path is app/Modules.
app/Modules/{ModuleName}/
├── App/ # PSR-4 source (capitalized)
│ ├── Console/Commands/ # auto-registered
│ ├── Http/{Controllers,Middleware,Requests}/
│ ├── Jobs/ Models/ Services/ Rules/ Enums/
├── Routes/ # auto-loaded
│ ├── api.php web.php console.php
├── Tests/{Unit,Integration,Component}/
└── agents/
Namespace: App\Modules\{ModuleName}\App\{Layer}\{Class} — note the
extra App segment.
Auto-loading: app/Providers/ModuleServiceProvider.php scans
app/Modules/ and registers routes (Routes/api.php → /api prefix +
api middleware; Routes/web.php → web middleware) plus Artisan
commands under App/Console/Commands/.
Route conventions:
// Routes/api.php — auto-prefixed /api
Route::name('v1.')->prefix('v1/{module-prefix}')->group(function(): void {
Route::get('/', [Controller::class, 'index'])->name('{module-prefix}.index');
});
Module name = PascalCase. Route prefix = kebab-case. Route names = dot notation.
Triggers when: modules.namespace_template starts with App\ (no
Modules segment) or root path is src/ with <Domain>/ subdirs.
src/{Domain}/
├── Application/ # use cases, command handlers
├── Domain/ # entities, value objects
├── Infrastructure/ # adapters, repositories
├── UserInterface/ # controllers, console
└── Tests/
Namespace: App\{Domain}\{Layer}\{Class} — no extra segment.
Auto-loading: Symfony service container auto-wires each
{Domain}/ subtree per services.yaml resource imports.
Triggers when: root path is packages/ and each child has
package.json.
packages/{pkg-name}/
├── package.json
├── src/ # entry points re-exported via "main"
├── tests/ __tests__/
└── README.md
Module identity comes from package.json#name, not the directory
name. The agent folder still lives at
packages/{pkg-name}/{modules.agent_folder}/.
Triggers when: root path is src/ and each child has __init__.py.
src/{package_name}/
├── __init__.py
├── <module files>
└── tests/ # or root-level tests/{package_name}/
Namespace is the import path: {package_name}.<sub>.<class>. Project
metadata in pyproject.toml.
Triggers when: root path is internal/ (one-level structure).
internal/{pkgname}/
├── *.go # package files
└── *_test.go # tests colocated
Import path: {module-path-from-go.mod}/internal/{pkgname}. No
top-level grouping — each subdir of internal/ is its own package.
{modules.agent_folder}/ scaffold.modules.namespace_template (when set).php artisan route:list for
Laravel HMVC, bin/console debug:container for Symfony,
npm test --workspace=<pkg> for Node monorepo).App segment
(App\Modules\X\App\…). Other stacks do not.modules.root_paths.packages/ no).<app>/<Modules>/, Symfony src/Domain, Node packages/, …) in
skill bodies, commands, or context docs — read from
modules.root_paths.framework: laravel frontmatter back to this skill —
the lint guard fails the build.まだレビューはありません。使ってみた感想をお寄せください。
概要と使いどころ
Use when reviewing UI for accessibility — WCAG 2.2 AA, keyboard nav, focus, ARIA, contrast, screen-reader semantics — even on 'is this a11y-OK?' or 'mach das barrierefrei'.
日本語の概要は準備中です。原文の説明を表示しています。
Use when defining or auditing the activation event — aha-moment selection, retention correlation, falsifiable definition. Triggers on 'what is our aha moment', 'redefine activation'.
日本語の概要は準備中です。原文の説明を表示しています。
Use when capturing an architectural decision — file naming, next ADR number, Status / Context / Decision / Consequences, index regen; fires even without saying 'ADR'.
日本語の概要は準備中です。原文の説明を表示しています。
Adversarial critique — devil's advocate, stress-test, honest teardown ('poke holes', 'be brutal', 'was hältst du davon'); explicit request only. Routine code or design review → code-review.
日本語の概要は準備中です。原文の説明を表示しています。
Use when reading, creating, or updating agent documentation, module docs, roadmaps, or AGENTS.md. Understands the full .augment/, agents/, and copilot-instructions structure.
日本語の概要は準備中です。原文の説明を表示しています。
Use for an adversarial red-team / blue-team / auditor review of an AI agent's CONFIG + behaviour (rules, skills, MCP, hooks, permissions) — attack-chain → defensive-gap list, not a code audit.
日本語の概要は準備中です。原文の説明を表示しています。