convez.ai
SkillDesenvolvimento WebIntermediário
dsiddharth2

PlugVault — Registro Oficial de Skills do Claude Code

dsiddharth2/plugvault

Registry de skills e comandos reutilizáveis para Claude Code

Ver no GitHub
InicialPluginScore · de 0 a 100

Registro centralizado e instalável de skills e comandos reutilizáveis para Claude Code, gerenciado via CLI plug. Inclui padrões de API, segurança, testes, Git, React, tratamento de erros e estilo de código, além de comandos para code review, geração de testes, refatoração e documentação.

Instalação

$npx convez-ai dsiddharth2/plugvault

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 1 dia
Versão
—
Issues abertas
0
Contribuidores
2

Casos de uso

  • Padronizar padrões de API REST em equipes
  • Aplicar guardrails de segurança automaticamente
  • Gerar testes e documentação com comandos Claude
  • Executar code reviews estruturados e refatorações
  • Compartilhar convenções de Git e React entre projetos

Veredito

PlugVault é uma solução pragmática para padronizar práticas de desenvolvimento em equipes que usam Claude Code, oferecendo skills e comandos prontos para revisar código, gerar testes e documentação. Cobre aspectos críticos (segurança, testes, convenções), mas depende do ecossistema plug CLI para funcionar e requer manutenção colaborativa do registry.

README

A tradução para o português deste README ainda está na fila.

PlugVault — Official Skills Registry

The official registry of reusable Claude Code skills and commands, installable via the plug CLI.

npm install -g plugvault
plug install -i code-review

What's Inside

Skills (always active in .claude/skills/)

Skills are background instructions that Claude follows automatically whenever they're relevant.

NameDescription
api-patternsREST API design — controller/service/repo layering, response envelopes, pagination, versioning
security-rulesSecurity guardrails — no hardcoded secrets, input validation, parameterized queries, auth checks
testing-standardsTesting conventions — Arrange-Act-Assert, naming, coverage targets, mock hygiene
git-conventionsGit workflow — conventional commits, branch naming, PR standards
react-patternsReact best practices — component structure, hooks, state management, effects
error-handlingError patterns — custom error classes, API error responses, structured logging, retries
code-styleCode style — naming conventions, file structure, imports, function design, comments

Commands (invoked with /command-name)

Commands are on-demand workflows you trigger explicitly.

NameDescription
code-reviewDeep code review across correctness, security, performance, maintainability
gen-testsGenerate comprehensive tests — happy path, edge cases, error cases
refactorAnalyze and refactor code — extract, simplify, rename, restructure
explain-codebaseWalk through project structure, architecture, entry points, conventions
gen-docsGenerate inline docs (JSDoc/TSDoc/docstrings) or markdown API references
pr-descriptionGenerate a structured PR description from the current branch's changes
debug-helpAnalyze an error/stack trace and provide diagnosis with fix

How It Works

Each entry in the registry is a folder containing:

registry/<name>/
├── meta.json       # Name, type, version, description, tags, entry filename
└── SKILL.md        # The actual skill file (for skills)
└── <name>.md       # The actual command file (for commands)

The plug CLI reads registry.json to discover available packages, then downloads the entry file and copies it to the correct location:

  • Skills → .claude/skills/<name>/SKILL.md (or ~/.claude/skills/ with -g)
  • Commands → .claude/commands/<name>.md (or ~/.claude/commands/ with -g)

File Formats

registry.json

Top-level index of all available packages:

{
  "name": "plugvault-official",
  "version": "1.0.0",
  "packages": {
    "api-patterns": {
      "type": "skill",
      "version": "1.0.0",
      "path": "registry/api-patterns",
      "description": "Enforces consistent REST API design patterns"
    }
  }
}

meta.json

Per-package metadata:

{
  "name": "api-patterns",
  "type": "skill",
  "version": "1.0.0",
  "description": "Enforces consistent REST API design patterns",
  "author": "plugvault",
  "tags": ["api", "rest", "architecture", "patterns"],
  "entry": "SKILL.md"
}

Skill/Command Files

All skill and command files use Claude Code's standard format with YAML frontmatter:

---
name: api-patterns
description: Enforces consistent REST API design patterns...
allowed-tools: Read, Grep, Glob
---

Your skill instructions here...

See Claude Code Skills Documentation for the full specification.

Contributing

Adding a New Skill

  1. Create a folder under registry/ with your skill name (kebab-case).
  2. Add meta.json with name, type ("skill"), version, description, author, tags, and entry ("SKILL.md").
  3. Add SKILL.md with YAML frontmatter (name, description) and your skill instructions.
  4. Add your skill to registry.json under packages.
  5. Open a PR.

Adding a New Command

  1. Create a folder under registry/ with your command name (kebab-case).
  2. Add meta.json with name, type ("command"), version, description, author, tags, and entry.
  3. Add <name>.md with YAML frontmatter (name, description, argument-hint, allowed-tools) and your command instructions.
  4. Add your command to registry.json under packages.
  5. Open a PR.

Guidelines

  • Skills should be general enough to apply to most projects in their domain.
  • Commands should produce structured, actionable output.
  • Keep descriptions concise — front-load the key use case.
  • Include practical, specific rules — not vague advice.
  • Test your skill/command in a real project before submitting.

Using a Private Vault

You can host your own registry with the same structure and register it with plug:

plug vault add personal https://github.com/yourname/my-vault --token ghp_xxx
plug install -i personal/deploy-checklist

License

MIT

Ver o original no GitHub

Bom para

Relacionados