القائمة
Auditoria completa de software (wrapper FORTRESS). Roda MERIDIAN + SENTINEL + ARCHITECT + CONDUCTOR + LIGHTHOUSE + HARNESS (opcional) em sequencia. Fortress Score = laudo completo.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Use quando Codex, Claude Code, outro agente, plataforma ou automacao precisar consultar o Data Engine Knowledge Gateway e pedir acesso governado a dados/APIs via Access Broker CLI. Cobre descoberta de recursos, manifesto, validacao, preview, request, status, drift e descriptor, sem aprovar ou revelar segredos.
Entry point do sistema. Recebe qualquer pedido, classifica, roteia para a melhor combinação de skills/agents/plugins, e monitora. Vai além do óbvio — sugere combos compostos, ativa auxiliares proativos, e delega a plugins canonicals (ADR-0001) quando apropriado.
Maquina especialista em dados Raiz: data-engine API canonical (/v1/kpis/*, /reports/<panel>/data, Envelope v2, KPI Registry Ouro M01..A*) + SQL TOTVS RM + Neon PostgreSQL + Zeev BPM + PBI_RAIZ. Use para: descobrir KPI Ouro, consumir painel canonical, escrever query nova (cascade painel>KPI>dinamica>raw), otimizar SQL lenta, anti-patterns, relatorios, processos Zeev. Tier 0: data-engine. Aplica ADR-041 (TOTVS via Neon Mirror enforced).
Maquina de design UI/UX. 67 estilos, 96 paletas, 57 font pairings, 13 stacks. Design system, componentes, landing pages, dashboards. Integra com shadcn, Figma, v0.
Maquina autonoma de limpeza de dead code (Knip + AST + bundle, confidence tiers, PRs atomicos). Use para tech debt, hand-off, reducao de bundle, cleanup pos-spike, auditoria de saude. Modo --target=harness audita dead skills/hooks/MCPs do proprio Claude Code.
Arquiteto senior platform + AI-first + data + LGPD para o ecossistema Raiz Educacao. Nao validador — par tecnico que desafia, traz benchmark externo, propoe alternativas, mede trade-offs. 10 modos de operacao. Carrega base do manual + state atual da Raiz como contexto.
| name | ag-9-auditar |
| description | Auditoria completa de software (wrapper FORTRESS). Roda MERIDIAN + SENTINEL + ARCHITECT + CONDUCTOR + LIGHTHOUSE + HARNESS (opcional) em sequencia. Fortress Score = laudo completo. |
| model | sonnet |
| context | fork |
| argument-hint | [URL ou path] [--resume] [--include-harness] [--harness-only] |
| allowed-tools | Read, Write, Edit, Glob, Grep, Bash, Agent, TaskCreate, TaskUpdate, TaskList, TeamCreate, TeamDelete, SendMessage |
| metadata | {"filePattern":"fortress-state.json,fortress-*.md,harness-audit-*.md","bashPattern":"auditar|fortress|audit","priority":93} |
/auditar https://app.example.com # FORTRESS classico (5D)
/auditar ~/Claude/GitHub/raiz-platform # FORTRESS classico
/auditar --resume # Retoma de fortress-state.json
/auditar --include-harness ~/Claude/... # FORTRESS + HARNESS (6D, default mensal)
/auditar --harness-only # So auditoria do harness local (sem app target)
Delega para ag-fortress (FORTRESS) — roda 5 ou 6 maquinas em sequencia:
--include-harness ou --harness-onlySem HARNESS (default):
Fortress Score (FS) = MQS*0.25 + SSS*0.25 + AQS*0.20 + DXS*0.15 + OBS*0.15
Com HARNESS (--include-harness):
FS6 = MQS*0.22 + SSS*0.22 + AQS*0.18 + DXS*0.13 + OBS*0.13 + HCS*0.12
Convergencia: FS >= 80 = passa. FS < 60 = bloqueio P0.
--include-harness--harness-onlydocs/diagnosticos/harness-audit-YYYY-MM-DD.md~/Claude/.claude/state/harness-audit-history.jsonl (HARNESS history)# Default (5D)
Agent({
subagent_type: "ag-fortress",
prompt: "[input completo do usuario]",
model: "opus",
run_in_background: true
})
# Com HARNESS (6D)
Agent({
subagent_type: "ag-fortress",
prompt: "[input do usuario] --include-harness",
model: "opus",
run_in_background: true
})
# So HARNESS
Skill({
skill: "ag-auditar-harness",
args: "$ARGUMENTS"
})
harness-coverage.md — R9 exige HCS >= 80 em PR que toca .claude/