# 应用工坊 · App Studio
Generate or standardize Lovstudio.AI apps, choosing web-only, PWA, Tauri,
or native macOS case-by-case. It supports Finder Quick Actions as signed Action
Extensions, React/Vite or Next.js, shadcn/ui, TanStack Query when useful,
Lovstudio.AI branding, CI/CD/deploy, optional auto update, and lovinsp. Every generated
or standardized frontend app runs the idempotent `lov-integrate-lovinsp` workflow by
default.
Independent source repository, also distributed through [lovstudio/skills](https://github.com/lovstudio/skills) — by [Lovstudio.AI](https://lovstudio.ai)
## Install
```bash
npx skills add lov-app-generator -g -y
```
Or through Claude Code plugin marketplace:
```text
/plugin marketplace add lovstudio/skills
/plugin install dev-tools@lovstudio
```
Requires: Python 3.8+ for the audit helper. No Python packages are required.
## Usage
```bash
# Ask the assistant:
生成一个 Lovstudio.AI Tauri App，品牌用 Lovstudio.AI，包含 shadcn、TanStack Query、CI/CD、自动更新和 lovinsp
# Or create web-only when desktop packaging is not needed:
生成一个只创建 web 的 Lovstudio.AI App，按需求判断用 Vite 还是 Next.js，包含 Warm Academic、shadcn 和 lovinsp
# Or require a real Finder Quick Action rather than a Services fallback:
生成一个原生 macOS App，对 Finder 选中的图片执行压缩；必须显示在 Quick Actions 和预览面板，不要做成 Services
# Or audit an existing app:
python3 "$SKILL_APP_GENERATOR_SKILL_DIR/scripts/audit_app_project.py" --root . --app-type auto --format markdown
```
## What It Does
1. Collects the app brief: name, slug, app type, platform, screens, backend, and release/deploy channel.
2. Audits the target project for Lovstudio.AI app requirements.
3. Chooses web-only, PWA, Tauri desktop, or native macOS from the brief instead of forcing one shell.
4. Guides new app scaffolding or incremental upgrade.
5. Applies the Warm Academic UI system and Lovstudio.AI brand asset paths.
6. Always invokes `lov-integrate-lovinsp` for frontend apps, including version
updates and migration from supported `code-inspector` integrations.
7. Coordinates other related Lovstudio.AI skills:
`install-shadcn-ui`, `install-tanstack-query`, `install-tauri-logo`,
and `project-port`.
8. Adds or checks CI/CD, web deploy wiring, and Tauri updater wiring when applicable.
9. Runs the lightest reliable verification commands available in the project.
10. Treats Finder Quick Actions as a distinct delivery surface, rejects Service-only
substitutes, and verifies the embedded Action Extension configuration.
## Options
| Option | Default | Description |
|--------|---------|-------------|
| `--root` | `.` | Target app root to inspect |
| `--app-type` | `auto` | Audit profile: `auto`, `web`, `tauri`, or `macos` |
| `--native-integration` | `auto` | Native surface: `auto`, `none`, or `finder-quick-action` |
| `--format` | `markdown` | Output format: `markdown` or `json` |
| `--output` | stdout | Optional path to write the audit report |
## User Configuration
Prefer environment variables when local paths differ:
| Variable | Usage |
|---|---|
| `SKILL_APP_GENERATOR_SKILL_DIR` | Installed `lov-app-generator` skill directory |
| `SKILL_DESIGN_GUIDE` | Warm Academic design guide path |
| `SKILL_PROFILE_PATH` | Lovstudio.AI brand asset root or profile |
## Brand Configuration
No personal path is built into this repository. Resolve brand assets through
explicit paths, `SKILL_PROFILE_PATH`,
`SKILL_DESIGN_GUIDE`, or the shared profile at
`${SKILL_PROFILE_PATH:-$HOME/.lovstudio/skills/profile.json}`.
See `references/user-config.md` for the complete resolution order.
## Audit Helper
```bash
python3 "$SKILL_APP_GENERATOR_SKILL_DIR/scripts/audit_app_project.py" --root /path/to/app --app-type auto
python3 "$SKILL_APP_GENERATOR_SKILL_DIR/scripts/audit_app_project.py" --root /path/to/app --app-type web --format json
python3 "$SKILL_APP_GENERATOR_SKILL_DIR/scripts/audit_app_project.py" --root /path/to/app --app-type tauri --native-integration finder-quick-action
python3 "$SKILL_APP_GENERATOR_SKILL_DIR/scripts/audit_app_project.py" --root /path/to/native-app --app-type macos --native-integration finder-quick-action
```
## License
MIT