RAG System Architect
Dennis0424/claude-skills
Arquiteta e implementa sistemas RAG em produção com diagnóstico automático
Ver no GitHubSkill que guia Claude na arquitetura e construção de sistemas RAG (Retrieval-Augmented Generation) em produção. Realiza diagnóstico automático dos dados, prescreve técnicas apropriadas em 4 camadas (chunking, query enhancement, retrieval, evaluation) e gera plano de implementação concreto com 37+ técnicas documentadas.
Instalação
npx convez-ai Dennis0424/claude-skills1 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
Casos de uso
- Projetar pipelines de busca semântica e QA sobre documentos
- Otimizar sistemas de retrieval para grandes volumes de dados
- Evaluar e comparar técnicas RAG para casos de uso específicos
Veredito

Skill bem estruturada que sistematiza decisões complexas de arquitetura RAG através de diagnóstico guiado. Útil para developers que precisam implementar busca semântica em produção, mas requer conhecimento prévio de conceitos RAG para máximo aproveitamento.
README
A tradução para o português deste README ainda está na fila.
claude-skills
Personal Claude Code skills — reusable technique guides for AI-assisted development.
Skills
rag-system-architect
A skill that guides Claude to architect and build any production-grade RAG system. It diagnoses your data and requirements through a short interview (or by scanning your project files automatically), prescribes a technique stack, and generates a concrete implementation plan.
Phases:
- Diagnose — auto-detect your data files, or answer 5 questions about your data type, volume, query patterns, and constraints
- Prescribe — recommends techniques across 4 layers: chunking, query enhancement, retrieval, and evaluation
- Implement — generates a full implementation plan with file structure, component breakdown, and framework mappings
Supported techniques: 37 production RAG techniques across chunking, query enhancement, retrieval, and evaluation layers. See technique-matrix.md for the full reference.
Credits
The RAG techniques referenced in this skill are sourced from:
NirDiamant/RAG_Techniques by Nir Diamant
A comprehensive collection of advanced Retrieval-Augmented Generation (RAG) techniques, implemented as Jupyter notebooks with LangChain and LlamaIndex. Covers 37+ techniques from foundational to advanced architectures including Graph RAG, RAPTOR, Self-RAG, CRAG, and more.
All notebook links in this skill point directly to that repository. Full credit to Nir Diamant and contributors for the original implementations.
Usage
These skills are loaded automatically by Claude Code from ~/.claude/skills/. Invoke with /rag-system-architect or Claude will load the skill automatically when you mention RAG, vector search, document QA, semantic search, or similar topics.