mega-brain-ai 1.3.0

package/.antigravity/README.md

@@ -0,0 +1,46 @@
+# 🎛️ .antigravity/
+
+> **Sistema de Extensões e Overrides Locais**
+
+Esta pasta permite customizações locais sem modificar o core do Mega Brain.
+
+## Estrutura
+
+```
+.antigravity/
+└── rules/          # Regras customizadas (override do .claude/rules/)
+    └── .gitkeep
+```
+
+## Como Usar
+
+### Adicionar Override de Regra
+
+Crie um arquivo `.md` em `rules/` com o mesmo nome da regra original:
+
+```markdown
+# Override: RULE-GROUP-1.md
+
+> Este arquivo sobrescreve comportamentos específicos do RULE-GROUP-1
+
+## Modificações Locais
+
+- [Suas customizações aqui]
+```
+
+### Prioridade de Carregamento
+
+1. `.claude/rules/*.md` (regras base)
+2. `.antigravity/rules/*.md` (overrides - maior prioridade)
+
+## Padrão Aios-Core
+
+Este sistema é inspirado no padrão `.antigravity/` do [aios-core](https://github.com/oalanicolas/aios-core), permitindo extensões sem poluir o repositório principal.
+
+## Gitignore
+
+Por padrão, o conteúdo de `.antigravity/rules/` é **trackeado** no git. Se quiser regras locais não commitadas, adicione ao `.gitignore`:
+
+```gitignore
+.antigravity/rules/*.local.md
+```

package/.antigravity/rules/mega-brain.md

@@ -0,0 +1,39 @@
+# Mega Brain - AI Knowledge Management System
+
+## Project Identity
+Mega Brain is an AI-powered knowledge management system that transforms expert materials (videos, PDFs, transcriptions) into structured playbooks, DNA schemas, and mind-clone agents.
+
+## Architecture
+```
+inbox/          -> Raw materials (videos, PDFs, transcriptions)
+processing/     -> Pipeline stages (chunks, canonical, insights, narratives)
+knowledge/      -> Structured knowledge base (dossiers, playbooks, DNA)
+agents/         -> AI agents (persons, cargo, council)
+system/         -> Protocols, schemas, documentation
+.claude/        -> Claude Code integration (hooks, skills, commands)
+bin/            -> CLI tools and installer
+```
+
+## DNA Schema (5 Knowledge Layers)
+| Layer | Name | Description |
+|-------|------|-------------|
+| L1 | PHILOSOPHIES | Core beliefs and worldview |
+| L2 | MENTAL-MODELS | Thinking and decision frameworks |
+| L3 | HEURISTICS | Practical rules and decision shortcuts |
+| L4 | FRAMEWORKS | Structured methodologies and processes |
+| L5 | METHODOLOGIES | Step-by-step implementations |
+
+## Non-Negotiable Rules
+1. All credentials in `.env` only — never hardcode API keys
+2. Python hooks use `pathlib.Path` for cross-platform compatibility
+3. Follow naming conventions: folders=lowercase, configs=SCREAMING-CASE, scripts=snake_case
+4. `.env` is gitignored — never commit credentials
+5. Run `npx mega-brain-ai setup` for initial configuration
+
+## Configuration
+- `.env` is the ONLY source of truth for credentials
+- Required: `OPENAI_API_KEY` (Whisper transcription)
+- Recommended: `VOYAGE_API_KEY` (semantic search)
+- Optional: `GOOGLE_CLIENT_ID` + `GOOGLE_CLIENT_SECRET` (Drive import)
+
+## For full documentation, see `.claude/CLAUDE.md`

package/.claude/CLAUDE.md

@@ -0,0 +1,172 @@
+# Mega Brain - AI Knowledge Management System
+
+## What is Mega Brain?
+
+AI-powered system that transforms expert materials (videos, PDFs, transcriptions) into structured playbooks, DNA schemas, and mind-clone agents. Powered by JARVIS orchestrator.
+
+## Quick Start
+
+1. Run `npx mega-brain-ai setup` (auto-triggers on first use if `.env` missing)
+2. Fill in API keys when prompted (only `OPENAI_API_KEY` is required)
+3. Use `/jarvis-briefing` to see system status
+
+## Architecture
+
+```
+mega-brain/
+├── core/           -> Engine (Pedro pattern)
+│   ├── tasks/          -> Atomic tasks (HO-TP-001 anatomy)
+│   ├── workflows/      -> YAML orchestration
+│   ├── intelligence/   -> Python scripts
+│   ├── patterns/       -> YAML configs
+│   ├── protocols/      -> Pipeline, conclave, DNA protocols
+│   ├── schemas/        -> JSON schemas
+│   ├── jarvis/         -> JARVIS Soul + DNA
+│   ├── templates/      -> Log templates
+│   └── glossary/       -> Domain glossaries
+├── agents/         -> AI agents (conclave, cargo, persons)
+├── .claude/        -> Claude Code integration
+├── docs/           -> Documentation, PRDs, plans
+│   ├── prd/            -> Product requirements
+│   └── plans/          -> Plan mode outputs
+├── bin/            -> CLI tools (npm)
+├── inbox/          -> Raw materials (L3)
+├── artifacts/      -> Pipeline stages (L3)
+├── knowledge/      -> Knowledge base (L3)
+└── logs/           -> Session logs (L3)
+```
+
+## Plan Mode
+
+Plans MUST be saved to `docs/plans/` (not ~/.claude/plans/).
+When in plan mode, save the plan file to: `docs/plans/YYYY-MM-DD-description.md`
+
+### Layer System
+
+| Layer | Content | Git Status |
+|-------|---------|------------|
+| L1 (Community) | core/, agents/conclave, .claude/, bin/, docs/ | Tracked (npm package) |
+| L2 (Pro) | agents/cargo, agents/sub-agents | Tracked (premium) |
+| L3 (Personal) | .data/, .env, agents/persons | Gitignored |
+
+## Community vs Pro
+
+| Feature | Community | Pro |
+|---------|-----------|-----|
+| CLI & Templates | yes | yes |
+| Skills & Hooks | yes | yes |
+| Agent Templates | yes | yes |
+| Knowledge Base (populated) | - | yes |
+| Mind Clone Agents | - | yes |
+| Pipeline Processing | - | yes |
+| Council / Conclave | - | yes |
+
+## DNA Schema (5 Knowledge Layers)
+
+| Layer | Name | Description |
+|-------|------|-------------|
+| L1 | PHILOSOPHIES | Core beliefs and worldview |
+| L2 | MENTAL-MODELS | Thinking and decision frameworks |
+| L3 | HEURISTICS | Practical rules and decision shortcuts |
+| L4 | FRAMEWORKS | Structured methodologies and processes |
+| L5 | METHODOLOGIES | Step-by-step implementations |
+
+## Commands
+
+| Command | Description |
+|---------|-------------|
+| `/jarvis-briefing` | Operational status + health score |
+| `/jarvis-full` | Full pipeline (ingest + process + enrich) |
+| `/process-jarvis` | Pipeline processor (5 phases) |
+| `/conclave` | Council session (multi-agent debate) |
+| `/ingest` | Ingest new material |
+| `/save` | Save current session |
+| `/resume` | Resume previous session |
+| `/setup` | Environment setup wizard |
+
+## Agents
+
+Defined in `AGENT-INDEX.yaml`, activated via slash commands.
+
+| Type | Count | Purpose |
+|------|-------|---------|
+| CARGO | 29 | Functional roles (Sales, Marketing, Ops) |
+| MINDS | 5 | Expert mind clones |
+| CONCLAVE | 3 | Multi-perspective deliberation |
+| SYSTEM | 2 | JARVIS, Agent-Creator |
+
+**Total Active Agents:** 37
+## Configuration
+
+- **`.env`** is the ONLY source of truth for credentials
+- Run `/setup` to configure interactively
+- Never hardcode API keys anywhere
+- `.mcp.json` uses `${ENV_VAR}` syntax for MCP servers
+
+### Required Keys
+
+| Key | Purpose | Required? |
+|-----|---------|-----------|
+| `OPENAI_API_KEY` | Whisper transcription | Yes (pipeline needs it) |
+| `VOYAGE_API_KEY` | Semantic embeddings (RAG) | Recommended |
+| `GOOGLE_CLIENT_ID` | Drive import | Optional |
+| `ANTHROPIC_API_KEY` | N/A with Claude Code | Not needed |
+
+## Hooks System
+
+20+ active hooks in `.claude/hooks/` (Python 3, stdlib + PyYAML only).
+Configured in `settings.json` (distributed) and `settings.local.json` (local overrides).
+
+| Event | Key Hooks |
+|-------|-----------|
+| SessionStart | `session_start.py`, `inbox_age_alert.py`, `skill_indexer.py` |
+| UserPromptSubmit | `skill_router.py`, `quality_watchdog.py`, `memory_updater.py` |
+| PreToolUse | `creation_validator.py`, `claude_md_guard.py` |
+| PostToolUse | `post_tool_use.py`, `enforce_dual_location.py` |
+| Stop | `stop_hook_completeness.py`, `ralph_wiggum.py` |
+
+## Rules (Lazy Loading)
+
+Detailed rules are loaded on-demand via keyword matching from `.claude/rules/`:
+
+| Group | Topics | File |
+|-------|--------|------|
+| PHASE-MANAGEMENT | phases, pipeline, batch | RULE-GROUP-1.md |
+| PERSISTENCE | sessions, save, resume | RULE-GROUP-2.md |
+| OPERATIONS | parallel, templates, KPIs | RULE-GROUP-3.md |
+| PHASE-5 | agents, dossiers, cascading | RULE-GROUP-4.md |
+| VALIDATION | source-sync, integrity | RULE-GROUP-5.md |
+| AUTO-ROUTING | skills, sub-agents, GitHub | RULE-GROUP-6.md |
+
+## Security
+
+1. **NEVER** hardcode API keys or tokens in code
+2. **ALWAYS** use `.env` for credentials (gitignored)
+3. Google OAuth credentials via config file, not code
+4. `git push` is blocked by `settings.json` deny rules — delegate to @devops
+
+## Conventions
+
+- Folders: lowercase (`inbox`, `system`)
+- Config files: SCREAMING-CASE (`STATE.json`, `MEMORY.md`)
+- Python scripts: snake_case, use `pathlib.Path` for cross-platform paths
+- Skills: kebab-case directories (`knowledge-extraction/`)
+
+## Troubleshooting
+
+| Problem | Solution |
+|---------|----------|
+| "Hook failed" | Check Python 3 is in PATH |
+| ".env not found" | Run `npx mega-brain-ai setup` |
+| "Permission denied on git push" | By design — use branch + PR workflow |
+| Skills not auto-activating | Check `SKILL-INDEX.json` is generated on SessionStart |
+
+## Recent Changes
+
+See `system/docs/CHANGELOG-ARCHITECTURE.md` for architectural evolution history.
+
+## CLAUDE.md Policy
+
+- Only 2 CLAUDE.md files are valid: root `CLAUDE.md` and `.claude/CLAUDE.md` (this file)
+- NEVER create CLAUDE.md in data or code subdirectories
+- Agent memory lives in `.claude/jarvis/` and `.claude/skills/`

package/.claude/agents.yaml

@@ -0,0 +1,44 @@
+# Claude Code Agent Registry
+# Source: agents/persona-registry.yaml
+# Version: 1.0.0
+
+source: "agents/persona-registry.yaml"
+sync_targets:
+  - ".cursor/agents.yaml"
+  - ".windsurf/agents.yaml"
+
+# Layer paths
+paths:
+  L0: "core/jarvis/"
+  L1: "agents/conclave/"
+  L2: "agents/boardroom/"
+  L3: "agents/minds/"
+  L4: "agents/cargo/"
+  SUB: ".claude/jarvis/sub-agents/"
+
+# Activation
+activation:
+  # L0 - System
+  jarvis: "@jarvis"
+  agent-creator: "/create-agent"
+
+  # L1 - Conclave
+  conclave: "/conclave"
+
+  # L2-L4 - via /ask
+  ask: "/ask [agent-id]"
+
+# Creation triggers
+triggers:
+  role_detector:
+    script: "core/intelligence/role_detector.py"
+    thresholds:
+      L4: 5
+      L2: 10
+
+  pipeline:
+    phase: "5.2"
+    creates: "L3"
+
+  manual:
+    command: "/create-agent"

package/.claude/commands/agents.md

@@ -0,0 +1,161 @@
+---
+description: Mostra status dos agentes e suas MEMORYs
+---
+
+# AGENTS - Status dos Agentes
+
+> **Versão:** 1.0.0
+> **Alias:** `/a`
+
+---
+
+## SINTAXE
+
+```
+/agents [FLAGS]
+```
+
+| Flag | Descrição |
+|------|-----------|
+| (nenhuma) | Lista todos agentes e status |
+| `--outdated` | Lista agentes com MEMORY desatualizada (>3 dias) |
+| `--knowledge "pessoa"` | Quais agentes têm conhecimento sobre pessoa |
+| `--category SALES` | Filtra por categoria (SALES, C-LEVEL, OPERATIONS) |
+
+---
+
+## EXECUÇÃO
+
+### Step 1: Scan Agentes
+```
+SCAN /agents/ for AGENT-*.md and MEMORY-*.md files
+
+FOR each agent:
+  READ AGENT-*.md header for version, last_updated
+  READ MEMORY-*.md for:
+    - Last update date
+    - Persons mentioned
+    - Sources included
+  CALCULATE staleness (days since update)
+```
+
+### Step 2: Verificar Role-Tracking
+```
+READ /agents/DISCOVERY/role-tracking.md
+
+IDENTIFY:
+  - Roles at threshold (>=10 mentions) without agent
+  - Roles approaching threshold (>=5 mentions)
+```
+
+### Step 3: Gerar AGENTS STATUS
+```
+═══════════════════════════════════════════════════════════════════════════════
+                              AGENTS STATUS
+                         {TIMESTAMP}
+═══════════════════════════════════════════════════════════════════════════════
+
+🤖 AGENTES ATIVOS: {TOTAL_COUNT}
+
+C-LEVEL:
+   CRO      │ MEMORY atualizada: {DATE} │ Conhece: {N} pessoas
+   CFO      │ MEMORY atualizada: {DATE} │ Conhece: {N} pessoas
+   CMO      │ MEMORY atualizada: {DATE} │ Conhece: {N} pessoas
+   COO      │ MEMORY atualizada: {DATE} │ Conhece: {N} pessoas
+
+SALES:
+   CLOSER           │ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+   BDR              │ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+   SDS              │ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+   LNS              │ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+   SALES-MANAGER    │ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+   SALES-LEAD       │ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+   SALES-COORDINATOR│ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+   CUSTOMER-SUCCESS │ MEMORY: {DATE} │ {N} pessoas │ v{VERSION}
+
+───────────────────────────────────────────────────────────────────────────────
+
+⚠️ DESATUALIZADOS (>3 dias):
+   {AGENT_1}, {AGENT_2} precisam de refresh
+
+🚨 SUGERIDOS PARA CRIAÇÃO:
+   🔴 {ROLE_NAME} ({N} menções - threshold atingido)
+   🟡 {ROLE_NAME} ({N} menções - monitorar)
+
+═══════════════════════════════════════════════════════════════════════════════
+```
+
+---
+
+## OUTPUT COM --outdated
+
+```
+═══════════════════════════════════════════════════════════════════════════════
+                         AGENTES DESATUALIZADOS
+                         {TIMESTAMP}
+═══════════════════════════════════════════════════════════════════════════════
+
+⚠️ AGENTES COM MEMORY > 3 DIAS:
+
+   CFO
+   ├─ Última atualização: {DATE} ({N} dias atrás)
+   ├─ Conhece: {PERSONS_LIST}
+   └─ Faltam fontes: {SOURCES_NOT_INCLUDED}
+
+   COO
+   ├─ Última atualização: {DATE} ({N} dias atrás)
+   ├─ Conhece: {PERSONS_LIST}
+   └─ Faltam fontes: {SOURCES_NOT_INCLUDED}
+
+⭐️ AÇÃO SUGERIDA
+   Atualizar MEMORYs: executar /process-jarvis para novas fontes
+   Ou: sync manual das fontes faltantes
+
+═══════════════════════════════════════════════════════════════════════════════
+```
+
+---
+
+## OUTPUT COM --knowledge "pessoa"
+
+```
+═══════════════════════════════════════════════════════════════════════════════
+                    CONHECIMENTO SOBRE: {PESSOA}
+                         {TIMESTAMP}
+═══════════════════════════════════════════════════════════════════════════════
+
+📚 AGENTES QUE CONHECEM {PESSOA}:
+
+   CLOSER
+   ├─ Fontes: {SOURCE_IDS}
+   ├─ Insights: {N} (HIGH: {N}, MEDIUM: {N}, LOW: {N})
+   └─ Frameworks: {LIST}
+
+   CRO
+   ├─ Fontes: {SOURCE_IDS}
+   ├─ Insights: {N}
+   └─ Frameworks: {LIST}
+
+📄 DOSSIER DISPONÍVEL:
+   /knowledge/dossiers/persons/DOSSIER-{PESSOA}.md
+
+═══════════════════════════════════════════════════════════════════════════════
+```
+
+---
+
+## EXEMPLOS
+
+```bash
+# Ver todos agentes
+/agents
+
+# Ver desatualizados
+/agents --outdated
+
+# Quem conhece Cole Gordon?
+/agents --knowledge "Cole Gordon"
+
+# Apenas agentes de vendas
+/agents --category SALES
+```

package/.claude/commands/ask.md

@@ -0,0 +1,117 @@
+---
+description: Consulta agente (cargo ou persona) aplicando DNA Cognitivo
+argument-hint: [agente] [pergunta] - Ex: cro "Qual estrutura de time?" ou hormozi "Qual comissao?"
+---
+
+# /ask - Consulta com DNA Cognitivo
+
+## Descricao
+Consulta um agente especifico (cargo ou persona) sobre uma pergunta, aplicando DNA Cognitivo.
+
+## Uso
+```
+/ask [agente] [pergunta]
+```
+
+## Argumentos
+- `agente`: Nome do cargo (cro, cfo, closer, etc.) ou persona (hormozi, cole, etc.)
+- `pergunta`: A pergunta a ser respondida
+
+## Exemplos
+```
+/ask cro "Qual estrutura ideal de time de vendas para faturar R$1M/mes?"
+/ask hormozi "Qual comissao ideal para closers?"
+/ask closer "Como lidar com objecao de preco?"
+```
+
+---
+
+## INSTRUCOES DE EXECUCAO
+
+### SE agente e CARGO (cro, cfo, cmo, coo, closer, bdr, etc.):
+
+```
+1. CARREGAR:
+   a) agents/constitution/BASE-CONSTITUTION.md
+   b) /agents/cargo/{LEVEL}/{CARGO}/DNA-CONFIG.yaml (se existir)
+   c) /agents/cargo/{LEVEL}/{CARGO}/MEMORY.md (se existir)
+
+2. IDENTIFICAR DOMINIO da pergunta
+   Mapear para: vendas, hiring, compensation, scaling, operations, etc.
+
+3. CARREGAR DNA relevante:
+   - De /knowledge/dna/AGGREGATED/AGG-{DOMINIO}.yaml (se existir)
+   - De /knowledge/dna/persons/*/HEURISTICAS.yaml (filtrado por dominio)
+   - Priorizar itens com peso >= 0.70
+
+4. APLICAR CASCATA DE RACIOCINIO:
+   METODOLOGIA → FRAMEWORK → HEURISTICA → MODELO MENTAL → FILOSOFIA
+   (Ver core/templates/agents/reasoning-model.md)
+
+5. RESPONDER no formato:
+
+[COMO {CARGO}]
+
+{Posicao clara em 2-3 frases}
+
+RACIOCINIO:
+{Qual camada de DNA usou e como}
+
+EVIDENCIAS:
+• {ID}: "{citacao resumida}"
+• {ID}: "{citacao resumida}"
+
+CONFIANCA: {0-100}%
+{Justificativa}
+
+LIMITACOES:
+• {O que nao sei}
+• {Premissas assumidas}
+```
+
+### SE agente e PERSONA (hormozi, cole, sam, jordan, etc.):
+
+```
+1. IDENTIFICAR pessoa canonica:
+   - hormozi → ALEX-HORMOZI
+   - cole → COLE-GORDON
+   - sam → SAM-OVEN
+   - jordan → JORDAN-LEE
+
+2. CARREGAR DNA da pessoa:
+   /knowledge/dna/persons/{PESSOA}/
+   - FILOSOFIAS.yaml
+   - MODELOS-MENTAIS.yaml
+   - HEURISTICAS.yaml
+   - FRAMEWORKS.yaml
+   - METODOLOGIAS.yaml
+   - CONFIG.yaml
+
+3. SE DNA nao existe, carregar DOSSIER:
+   /knowledge/dossiers/persons/DOSSIER-{PESSOA}.md
+
+4. RESPONDER como se fosse a pessoa:
+   - Usar tom caracteristico (de CONFIG.yaml ou DOSSIER)
+   - Citar evidencias com IDs quando disponivel
+   - Manter perspectiva da pessoa
+
+5. FORMATO:
+
+[COMO {PESSOA}]
+
+{Resposta no tom e estilo da pessoa}
+
+BASEADO EM:
+• {ID ou trecho do DOSSIER}: "{citacao}"
+
+CONFIANCA: {0-100}%
+```
+
+---
+
+## NOTAS
+
+- Se nao houver DNA extraido para a pessoa, usar DOSSIER
+- Se nao houver DOSSIER, declarar que nao tem conhecimento dessa pessoa
+- Sempre citar fontes quando disponivel
+- Nunca inventar IDs ou evidencias