DeepArchi — Framework Multi-Agent para Arquitetura Empresarial
deeparchi-ai/deeparchi-skill
Multi-agent enterprise architecture com 9 agentes, DAG, validação cruzada e ArchiMate
Ver no GitHubFramework completo para design, implantação e operação de sistemas de arquitetura empresarial com múltiplos agentes IA. Oferece orquestração de 9 agentes especializados, validação cruzada dual-modelo, suporte ArchiMate 3.2 e governança estruturada via topologia DAG. Projetado para AI agents consumirem especificações de forma determinística.
Instalação
npx convez-ai deeparchi-ai/deeparchi-skill1 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
- Design de arquitetura empresarial multi-agente com governança explícita
- Orquestração de equipes de agentes IA especializados em diferentes domínios
- Modelagem arquitetural com ArchiMate 3.2 e draw.io integrado
- Validação de topologia, ciclos e blast radius em sistemas distribuídos
- Mapeamento entre TOGAF ADM e estrutura DeepArchi
Veredito

Framework robusto e bem documentado para orquestração de múltiplos agentes IA em contexto empresarial, com suporte a modelagem arquitetural padronizada. Requer expertise em arquitetura e governança para aproveitar plenamente o potencial.
README
A tradução para o português deste README ainda está na fila.
DeepArchi — Multi-Agent Enterprise Architecture Framework (MAEA)
Core metaphor: Agents are employees. Manage them like you manage people — but every hire is an AI.
DeepArchi is a comprehensive framework for designing, deploying, and operating multi-agent enterprise architecture systems. 9 specialized agents across 6 model providers, cross-validated for quality, operating at ~¥20,900/month.
What's inside
MAEA Framework
- 9-Agent organization — Architecture Guardian, Strategy, Product, Engineering, Data, Operations, Marketing, Risk & Compliance, Capital
- Topology DAG — Golden solid lines (governance), Cyan solid lines (task orchestration), Gray dashed lines (collaboration)
- Budget matrix — ¥20,900/month, per-agent breakdown with cross-validation costs
- Six design principles — Dual-model cross-validation, Agent-as-employee, Reversibility-first, Cross-vendor redundancy, Quantified confidence, Mandatory verification layer
- Human charter — Five immutable rules (mission, scope, prohibitions, capital structure, termination)
- Security Gene — Six hardcoded rules at the gateway layer
ArchiMate Modeling
Built-in ArchiMate 3.2 support with draw.io integration — four-layer architecture (Business, Application, Data, Technology) with color-coded elements.
Reference Deep-Dives
- Architecture Agent Design — Topology validator, onboarding checklist, escalation protocol
- Agent Designer Methodology — 9-phase agent design process
- Topology Validator — DAG consistency, cycle detection, blast radius
- TOGAF Mapping — Full TOGAF ADM ↔ DeepArchi mapping
- ArchiMate Modeling — ArchiMate 3.2 with draw.io
Quick start
# Load the skill in your AI agent
skill_view(name='deeparchi')
The framework is designed for AI agents to consume — every section is machine-actionable with explicit constraints, templates, and verification criteria.
Works with
- DeepSight — Deep reasoning research methodology
- DeepBLM — AI-era strategy consulting framework
License
MIT