Personal Claude Code plugin — reusable skills, curated plugin recommendations, and a custom statusline across all projects.
Run /plugins in Claude Code, search for hai@ng-hai, and enable it.
Add to ~/.claude/settings.json:
{
"extraKnownMarketplaces": {
"ng-hai": {
"source": {
"source": "git",
"url": "https://github.com/ng-hai/dotclaude.git",
"ref": "main"
}
}
},
"enabledPlugins": {
"hai@ng-hai": true
}
}Restart Claude Code, then run /setup for interactive configuration.
Remove the ng-hai entry from extraKnownMarketplaces and hai@ng-hai from enabledPlugins in ~/.claude/settings.json.
| Skill | Description |
|---|---|
craft |
8-phase feature development orchestrator — discovery, exploration, architecture, implementation, quality review, and ship |
| Skill | Description |
|---|---|
bulletproof-react |
Umbrella architectural blueprint — project structure, coding standards, references specialist skills |
| Skill | Description | Source |
|---|---|---|
react-best-practices |
55 React/Next.js performance rules across 7 categories | vercel-labs/agent-skills |
react-router-framework |
React Router framework mode — routes, loaders, actions, forms, pending/optimistic UI | remix-run/agent-skills |
composition-patterns |
8 composition rules across 4 categories — compound components, state lifting, React 19 APIs | vercel-labs/agent-skills |
tailwind-best-practices |
26 Tailwind CSS rules across 7 categories | — |
api-layer |
15 TanStack Query rules across 6 categories — data fetching, caching, optimistic updates | — |
state-management |
6 state rules across 6 categories — colocation, server vs client, Zustand, URL state | — |
error-handling |
5 error handling rules across 5 categories — boundaries, API interceptors, Sentry, degradation | — |
security |
5 frontend security rules across 3 categories — auth, XSS prevention, CSP | — |
testing-strategy |
6 testing rules across 4 categories — Vitest, Testing Library, MSW, Playwright | — |
web-design-guidelines |
UI audit for accessibility, performance, and UX | vercel-labs/agent-skills |
| Skill | Description |
|---|---|
react-doctor |
Scans React codebase for security, performance, correctness, and architecture issues (0-100 score) |
| Skill | Description |
|---|---|
setup |
Interactive wizard to configure plugins, statusline, and output styles |
Custom status bar showing worktree indicator, model name, context battery, and session cost.
The /setup skill recommends and configures official marketplace plugins with conflict detection. Run it to see what's available.
dotclaude/
├── .claude-plugin/
│ └── marketplace.json # Marketplace registration
├── skills/
│ ├── craft/ # 8-phase feature orchestrator
│ ├── bulletproof-react/ # Umbrella architecture blueprint
│ ├── react-best-practices/
│ ├── react-router-framework/
│ ├── composition-patterns/
│ ├── tailwind-best-practices/
│ ├── api-layer/
│ ├── state-management/
│ ├── error-handling/
│ ├── security/
│ ├── testing-strategy/
│ ├── web-design-guidelines/
│ ├── react-doctor/ # Diagnostic scanner
│ └── setup/ # Interactive setup wizard + registry
├── statusline.sh # Custom statusline renderer
└── README.md
Distributed directly from the main branch. Version must be bumped (semver) for changes to reach users — Claude Code caches plugins by version string.