convez.ai
SkillDesenvolvimento WebIntermediário
dopiotrek

dotclaude

dopiotrek/dotclaude

Framework agentic para Claude Code com hooks, agentes e skills

Ver no GitHub
InicialPluginScore · de 0 a 100

Framework configurável para Claude Code que fornece guardrails automatizados, agentes especializados, skills reutilizáveis e padrões sensatos para desenvolvimento assistido por IA. Inclui 13 hooks de validação/segurança, 9 agentes de domínio específico e configurações compartilháveis.

Instalação

$npx convez-ai dopiotrek/dotclaude

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 semana
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Automatizar validações de segurança e formatação de código
  • Executar tarefas especializadas (frontend, backend, debugging)
  • Compartilhar configurações e padrões de desenvolvimento em equipes
  • Monitorar e auditar operações do Claude Code

Veredito

Solução madura e bem estruturada para quem já usa Claude Code profissionalmente. Excelente para equipes que querem padronização, segurança e reutilização de configurações, mas requer familiaridade com hooks Python e fluxos de desenvolvimento ágil.

README

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

dotclaude

Claude Code agentic framework - hooks, agents, skills, and configuration

A comprehensive, shareable configuration framework for Claude Code that provides automated guardrails, specialized agents, reusable skills, and sensible defaults for productive AI-assisted development.

What's Included

ComponentCountDescription
Hooks13Automated guardrails for security, formatting, and validation
Agents9Specialized task handlers (frontend, backend, debugging, etc.)
Skills4Reusable skill definitions with reference materials
Settings1Comprehensive permission rules and hook configuration
CLAUDE.md1Global preferences and instructions

Quick Start

# Clone the repository
git clone https://github.com/dopiotrek/dotclaude.git ~/repos/dotclaude

# Run the installer (creates symlinks)
cd ~/repos/dotclaude
./install.sh

# Start a new Claude Code session
claude

The installer creates symlinks from ~/.claude to this repository, making it the single source of truth. Any changes you make here automatically apply to Claude Code.

Installation Options

./install.sh            # Default: symlink mode (recommended)
./install.sh --copy     # Copy files instead of symlinks
./install.sh --uninstall  # Remove symlinks and optionally restore backup

Hooks

Hooks are Python scripts that run before/after Claude Code tool invocations, providing automated guardrails:

Security & Safety

  • no-secrets.py - Blocks hardcoded API keys, tokens, and passwords
  • permission-auto-approve.py - Auto-approves safe read-only operations
  • bash-command-validator.py - Enforces use of modern CLI tools (rg over grep)

Code Quality

  • auto-format.py - Auto-formats code after edits (Prettier, Black, etc.)
  • import-path-validator.py - Enforces import conventions ($lib, package imports)
  • sveltekit-route-validator.py - Validates SvelteKit routing conventions
  • sveltekit-perf-guard.py - Monitors bundle size and performance anti-patterns

Database & Migrations

  • drizzle-migration-guard.py - Warns about destructive schema changes
  • supabase-rls-reminder.py - Reminds to add Row Level Security policies

Utilities

  • command-logger.py - Logs all Claude Code operations for audit
  • dependency-audit.py - Runs security audits when dependencies change
  • stop-verification.py - Runs type checks when Claude completes work
  • web-search-enhancer.py - Adds current year to tech documentation searches

See hooks/README.md for detailed documentation.

Agents

Specialized agents handle complex, domain-specific tasks:

AgentModelPurpose
frontend-engineersonnetSvelte 5 components, shadcn-svelte, responsive design
backend-engineersonnetSvelteKit server-side, load functions, form actions
superforms-expertsonnetsveltekit-superforms + Zod validation
debug-specialistsonnetError diagnosis, test failures, performance issues
code-revieweropusComprehensive security and quality review
discovery-agentsonnetRequirements gathering and feature specification
vercel-deployment-expertsonnetVercel deployment and configuration
seo-expertsonnetContent structure and SEO optimization
mobile-ui-designersonnetMobile-first UI design with TailwindCSS

See agents/README.md for detailed documentation.

Skills

Skills are reusable task definitions with supporting reference materials:

  • clean-comments - Remove unnecessary code comments
  • code-reviewer - Code review with checklists and analysis scripts
  • frontend-design - UI constraints for consistent design
  • svelte-component-architecture - Component decomposition guidelines

See skills/README.md for detailed documentation.

Configuration

CLAUDE.md

Global preferences and instructions that apply to all Claude Code sessions:

# Global Claude Preferences

## About Me

- Primary stack: SvelteKit, TypeScript, Supabase
- All projects use pnpm (never npm or yarn)

## Things to Always Do

- Run type checks before considering work complete
- Preserve existing code style in files being edited
- Use existing patterns from the codebase
  ...

Settings

The settings/settings.template.json contains:

  • Permissions - Fine-grained allow/deny/ask rules
  • Hooks - Hook configuration for each event type
  • Model - Default model preference
  • Plugins - Enabled marketplace plugins

Customization

Adding Your Own Hooks

  1. Create a Python script in hooks/:
#!/usr/bin/env python3
import json
import sys

def main():
    input_data = json.load(sys.stdin)
    tool_name = input_data.get("tool_name", "")

    # Your logic here

    sys.exit(0)  # 0=allow, 1=warn, 2=block

if __name__ == "__main__":
    main()
  1. Add the hook to settings/settings.template.json
  2. Re-run ./install.sh to regenerate settings

Adding Your Own Agents

Create a markdown file in agents/ with YAML frontmatter:

---
name: my-agent
description: When and how to use this agent
model: sonnet
color: blue
---

You are an expert in...

Modifying Skills

Skills in skills/ follow the structure:

skill-name/
├── SKILL.md           # Main skill definition
├── references/        # Supporting documentation
└── scripts/           # Automation scripts (optional)

Project Structure

dotclaude/
├── README.md                    # This file
├── CLAUDE.md                    # Global preferences
├── install.sh                   # Installation script
├── settings/
│   └── settings.template.json   # Settings with $HOME placeholders
├── hooks/
│   ├── README.md
│   └── *.py                     # 13 hook scripts
├── agents/
│   ├── README.md
│   └── *.md                     # 9 agent definitions
├── skills/
│   ├── README.md
│   └── */                       # 4 skill directories
└── docs/
    └── *.md                     # Additional documentation

Requirements

Optional Dependencies

Some hooks have optional dependencies for full functionality:

  • prettier - For auto-formatting JS/TS/Svelte
  • black - For auto-formatting Python
  • terminal-notifier (macOS) - For notifications

License

MIT License - Feel free to use, modify, and share.

Contributing

  1. Fork the repository
  2. Create a feature branch
  3. Make your changes
  4. Submit a pull request

Ideas for contributions:

  • New hooks for additional validations
  • Agents for other frameworks/languages
  • Skills for common development tasks
  • Documentation improvements
Ver o original no GitHub

Bom para

Relacionados