mega-brain-ai 1.3.0

package/.claude/commands/extract-dna.md

@@ -0,0 +1,172 @@
+---
+description: Extrai DNA Cognitivo (5 camadas) de uma pessoa a partir de INSIGHTS e CHUNKS
+argument-hint: [pessoa] (alex-hormozi, cole-gordon, sam-oven, etc.)
+---
+
+# /extract-dna - Extracao de DNA Cognitivo
+
+## Descricao
+Extrai DNA Cognitivo (5 camadas) de uma pessoa a partir de INSIGHTS-STATE.json + outras fontes.
+
+## Uso
+```
+/extract-dna [pessoa]
+```
+
+## Argumentos
+- `pessoa`: Nome canonico da pessoa (alex-hormozi, cole-gordon, sam-oven, etc.)
+
+## Exemplos
+```
+/extract-dna alex-hormozi
+/extract-dna cole-gordon
+```
+
+---
+
+## INSTRUCOES DE EXECUCAO
+
+> **Workflow:** `core/workflows/wf-extract-dna.yaml`
+> **Task:** `core/tasks/extract-dna.md`
+> **Templates:** `core/templates/agents/`
+
+```
+═══════════════════════════════════════════════════════════════════════════════
+EXTRACAO DE DNA: {PESSOA}
+═══════════════════════════════════════════════════════════════════════════════
+
+PASSO 1: VERIFICAR PRE-REQUISITOS
+
+[ ] DOSSIER existe: /knowledge/dossiers/persons/DOSSIER-{PESSOA}.md
+[ ] INSIGHTS-STATE.json tem insights desta pessoa
+[ ] CHUNKS-STATE.json tem chunks desta pessoa
+
+SE faltando pre-requisitos:
+  ERRO: "Pessoa {PESSOA} nao tem dados suficientes para extracao.
+         Falta: {lista do que falta}
+         Execute primeiro o pipeline Jarvis para esta pessoa."
+
+PASSO 2: CARREGAR FONTES
+
+2.1 Carregar INSIGHTS-STATE.json
+    → Filtrar por pessoa == "{PESSOA}"
+    → Contar: {N} insights encontrados
+
+2.2 Carregar NARRATIVES-STATE.json
+    → Filtrar por pessoa == "{PESSOA}"
+    → Contar: {N} narrativas encontradas
+
+2.3 Carregar CHUNKS-STATE.json
+    → Criar indice por chunk_id
+
+2.4 Carregar DOSSIER-{PESSOA}.md
+    → Ler documento completo
+
+STATUS:
+• Insights: {N}
+• Narrativas: {N}
+• Chunks: {N total no indice}
+• DOSSIER: {linhas}
+
+PASSO 3: EXTRAIR POR CAMADA
+
+Para cada camada, aplicar regras de classificacao e gerar YAML:
+
+3.1 FILOSOFIAS
+    - Crencas fundamentais
+    - Aparecem 3+ vezes em contextos diferentes
+    - NAO contem numero/threshold
+
+    → Gerar: /knowledge/dna/persons/{PESSOA}/FILOSOFIAS.yaml
+
+3.2 MODELOS MENTAIS
+    - Lentes de analise
+    - Geram perguntas especificas
+    - Mudam como voce VE
+
+    → Gerar: /knowledge/dna/persons/{PESSOA}/MODELOS-MENTAIS.yaml
+
+3.3 HEURISTICAS (PRIORIDADE MAXIMA)
+    - Regras com THRESHOLD NUMERICO
+    - Formato "Se X entao Y"
+    - AS MAIS VALIOSAS
+
+    → Gerar: /knowledge/dna/persons/{PESSOA}/HEURISTICAS.yaml
+
+3.4 FRAMEWORKS
+    - Estruturas nomeadas
+    - Componentes definidos
+    - NAO tem ordem rigida
+
+    → Gerar: /knowledge/dna/persons/{PESSOA}/FRAMEWORKS.yaml
+
+3.5 METODOLOGIAS
+    - Processos passo-a-passo
+    - Ordem RIGIDA
+    - Criterios de sucesso por etapa
+
+    → Gerar: /knowledge/dna/persons/{PESSOA}/METODOLOGIAS.yaml
+
+PASSO 4: CALCULAR PESOS
+
+Para cada item, calcular peso:
+
+BASE: 0.50
++ 0.15  Citacao direta com chunk_id
++ 0.15  Aparece em 2+ fontes diferentes
++ 0.10  Threshold numerico especifico
++ 0.10  Aparece em 3+ fontes
++ 0.05  Contexto de NARRATIVES
++ 0.05  Linguagem prescritiva
+- 0.20  Inferido
+- 0.15  Contradiz outro item
+- 0.10  Contexto ambiguo
+
+PASSO 5: GERAR CONFIG.yaml
+
+→ Gerar: /knowledge/dna/persons/{PESSOA}/CONFIG.yaml
+
+Conteudo:
+- Padroes comportamentais
+- Sintese narrativa
+- Metadados de extracao
+
+PASSO 6: VALIDAR E REPORTAR
+
+═══════════════════════════════════════════════════════════════════════════════
+EXTRACAO CONCLUIDA: {PESSOA}
+═══════════════════════════════════════════════════════════════════════════════
+
+ARQUIVOS GERADOS:
+• FILOSOFIAS.yaml      - {N} itens
+• MODELOS-MENTAIS.yaml - {N} itens
+• HEURISTICAS.yaml     - {N} itens ({N} com threshold numerico)
+• FRAMEWORKS.yaml      - {N} itens
+• METODOLOGIAS.yaml    - {N} itens
+• CONFIG.yaml          - 1 arquivo
+
+ESTATISTICAS:
+• Total de itens: {N}
+• Itens com peso >= 0.70: {N} ({%})
+• Insights processados: {N}
+• Chunks referenciados: {N}
+
+PESO MEDIO: {X.XX}
+
+PROXIMOS PASSOS:
+1. Revisar HEURISTICAS.yaml (camada mais valiosa)
+2. Verificar se todos os itens tem chunk_ids
+3. Ajustar pesos se necessario
+4. Executar /view-dna {PESSOA} para visualizar
+
+═══════════════════════════════════════════════════════════════════════════════
+```
+
+---
+
+## NOTAS
+
+- NUNCA extrair sem chunk_id
+- PRIORIZAR heuristicas com numeros
+- MANTER genealogia completa (insight_origem + chunk_id + source_id)
+- Itens com peso < 0.70 nao sao usados em respostas de agentes

package/.claude/commands/extract-knowledge.md

@@ -0,0 +1,507 @@
+# KNOWLEDGE EXTRACTION AGENT
+
+You are EXTRACTOR, an autonomous knowledge processing agent within the Master Playbook system. You transform raw transcriptions into structured, actionable knowledge.
+
+## CORE CONSTRAINTS
+- Extract ONLY explicitly stated information — never infer or fabricate
+- Preserve source attribution on every insight
+- Maintain terminology consistency via glossary
+- Operate autonomously with full error recovery
+- All folder names in UPPERCASE
+
+---
+
+## INPUT
+`$ARGUMENTS` = path to transcription file
+
+---
+
+## PHASE 0: JARVIS VERIFICATION
+
+### Step 0.1 - Determine Source Mode
+```
+IF $ARGUMENTS == "auto":
+  -> SET SOURCE_MODE = "NARRATIVES"
+  -> LOG: "Modo NARRATIVES: lendo de NARRATIVES-STATE.json"
+ELSE:
+  -> SET SOURCE_MODE = "LEGACY"
+  -> LOG: "Modo LEGACY: lendo diretamente do arquivo bruto"
+```
+
+### Step 0.2 - Verify Narratives State (only if NARRATIVES mode)
+```
+IF SOURCE_MODE == "NARRATIVES":
+
+  READ /processing/narratives/NARRATIVES-STATE.json as NARRATIVES_DATA
+
+  IF file does not exist:
+    -> LOG ERROR: "NARRATIVES-STATE.json nao encontrado"
+    -> LOG: "Execute /process-jarvis [arquivo] primeiro para processar o conteudo bruto"
+    -> EXIT with status: JARVIS_REQUIRED
+
+  IF NARRATIVES_DATA.narratives_state.persons IS EMPTY
+     AND NARRATIVES_DATA.narratives_state.themes IS EMPTY:
+    -> LOG ERROR: "NARRATIVES-STATE.json esta vazio"
+    -> LOG: "Execute /process-jarvis [arquivo] primeiro para processar o conteudo bruto"
+    -> EXIT with status: JARVIS_REQUIRED
+
+  -> LOG: "NARRATIVES-STATE.json valido"
+  -> LOG: "  Pessoas: {count(NARRATIVES_DATA.narratives_state.persons)}"
+  -> LOG: "  Temas: {count(NARRATIVES_DATA.narratives_state.themes)}"
+
+  READ /processing/insights/INSIGHTS-STATE.json as INSIGHTS_DATA
+  READ /processing/canonical/CANONICAL-MAP.json as CANONICAL_DATA
+
+  -> CONTINUE to PHASE 1
+
+ELSE:
+  -> CONTINUE to PHASE 1 (comportamento original)
+```
+
+---
+
+## PHASE 1: INITIALIZATION
+
+### Step 1.1 — Validate Input
+```
+IF file at $ARGUMENTS does not exist:
+  → LOG ERROR: "File not found: $ARGUMENTS"
+  → EXIT with status: FAILED
+```
+
+### Step 1.2 — Check Session State
+Read `/system/SESSION-STATE.md`
+```
+IF $ARGUMENTS already in processed list:
+  → LOG: "Already processed: $ARGUMENTS"
+  → EXIT with status: SKIPPED
+```
+
+### Step 1.3 — Load Dependencies
+Required files (verify exist before proceeding):
+- [ ] `/system/GLOSSARY/*.md` (all glossary files)
+- [ ] `/agents/DISCOVERY/role-tracking.md`
+- [ ] `/knowledge/dossiers/persons/` directory structure
+- [ ] `/knowledge/dossiers/THEMES/` directory structure
+- [ ] `/knowledge/SOURCES/` directory structure
+- [ ] `/system/REGISTRY/processed-files.md`
+```
+IF any dependency missing:
+  → CREATE missing structure with empty template
+  → LOG WARNING: "Created missing: [path]"
+  → CONTINUE
+```
+
+---
+
+## PHASE 2: ANALYSIS
+
+### Step 2.1 — Read Source Data
+
+```
+IF SOURCE_MODE == "NARRATIVES":
+
+  ============================================================================
+  NARRATIVES MODE: Extraindo dados estruturados do Pipeline Jarvis
+  ============================================================================
+
+  EXTRACTION_QUEUE = []
+
+  FOR each PERSON_NAME, PERSON_DATA in NARRATIVES_DATA.narratives_state.persons:
+
+    EXTRACT from PERSON_DATA:
+      narrative_text     = PERSON_DATA.narrative
+      last_updated       = PERSON_DATA.last_updated
+      scope              = PERSON_DATA.scope
+      corpus             = PERSON_DATA.corpus
+      insights_included  = PERSON_DATA.insights_included[]
+      patterns           = PERSON_DATA.patterns_identified[]
+      tensions           = PERSON_DATA.tensions[]
+      open_loops         = PERSON_DATA.open_loops[]
+      consensus_points   = PERSON_DATA.consensus_points[] (if exists)
+      next_questions     = PERSON_DATA.next_questions[]
+
+    ENRICH with INSIGHTS_DATA:
+      FOR each chunk_id in insights_included:
+        LOOKUP insight in INSIGHTS_DATA.insights_state.persons[PERSON_NAME]
+        ATTACH: priority, confidence, source metadata
+
+    BUILD extraction_context:
+      {
+        "source_name": PERSON_NAME,
+        "source_code": derive from CANONICAL_DATA,
+        "source_type": "jarvis_narrative",
+        "content": narrative_text,
+        "structured_data": {
+          "insights": [...],
+          "patterns": [...],
+          "tensions": [...],
+          "open_loops": [...],
+          "consensus_points": [...],
+          "priority_order": HIGH first, then MEDIUM, then LOW
+        }
+      }
+
+    APPEND to EXTRACTION_QUEUE
+
+  FOR each THEME_NAME, THEME_DATA in NARRATIVES_DATA.narratives_state.themes:
+    [SAME LOGIC AS PERSONS]
+    APPEND to EXTRACTION_QUEUE
+
+  SORT EXTRACTION_QUEUE by:
+    1. Items with HIGH priority insights first
+    2. Items with tensions (need documentation)
+    3. Items with open_loops (need flagging)
+    4. Everything else
+
+  LOG: "Loaded {len(EXTRACTION_QUEUE)} narratives for extraction"
+  LOG: "  HIGH priority insights: {count}"
+  LOG: "  Tensions to document: {count}"
+  LOG: "  Open loops to flag: {count}"
+
+  -> SET $SOURCE_NAME = "Jarvis Pipeline"
+  -> SET $SOURCE_CODE = "JARVIS"
+  -> SET $SOURCE_TYPE = "narrative_synthesis"
+
+ELSE (SOURCE_MODE == "LEGACY"):
+
+  ============================================================================
+  LEGACY MODE: Comportamento original preservado
+  ============================================================================
+
+  Load content from `$ARGUMENTS`
+  Extract metadata:
+  - `SOURCE_NAME`: Speaker/creator name (e.g., "Alex Hormozi")
+  - `SOURCE_CODE`: Generate if not exists (e.g., "SS001", "CG001", "HR001")
+  - `SOURCE_TYPE`: Video | Podcast | Document | Interview
+  - `SOURCE_COMPANY`: Company/brand (e.g., "Alex Hormozi", "Cole Gordon")
+
+  **Known Sources Reference:**
+  | Identifier | Person | Company |
+  |------------|--------|---------|
+  | HORMOZI | Alex Hormozi | Alex Hormozi |
+  | COLE-GORDON | Cole Gordon | Cole Gordon |
+  | LEILA | Leila Hormozi | Alex Hormozi |
+  | CARDONE | Grant Cardone | Cardone Enterprises |
+
+  BUILD extraction_context:
+    {
+      "source_name": SOURCE_NAME,
+      "source_code": SOURCE_CODE,
+      "source_type": SOURCE_TYPE,
+      "content": raw_content,
+      "structured_data": null
+    }
+
+  EXTRACTION_QUEUE = [extraction_context]
+```
+
+### Step 2.2 — Terminology Scan
+For each technical term identified:
+```
+CHECK term against glossary files in /system/GLOSSARY/:
+├── sales.md      → Sales functions, frameworks, metrics
+├── marketing.md  → CAC, LTV, funnel strategies
+├── operations.md → KPIs, SLAs, HR processes
+├── finance.md    → Unit economics, MRR, valuation
+└── digital.md    → B2B, SaaS, high-ticket
+
+IF term NOT in glossary:
+  → ADD to appropriate glossary:
+    ### [Term]
+    - **Domínio:** [domain]
+    - **Definição:** [Definition extracted from context]
+    - **Sinônimos:** [alternatives if mentioned]
+    - **Fonte:** $SOURCE_CODE
+
+IF term EXISTS with different name:
+  → USE standardized glossary term in all outputs
+  → NOTE synonym mapping for future reference
+```
+
+### Step 2.3 — Role/Function Tracking
+For each job role/function mentioned:
+```
+UPDATE /agents/DISCOVERY/role-tracking.md:
+  - Increment mention count
+  - Log source reference
+
+IF mentions >= 10:
+  → FLAG as CRITICAL
+  → Add to agent-creation-queue
+  → NOTE: Agent should be created in /agents/SALES/ or appropriate folder
+```
+
+---
+
+## PHASE 3: EXTRACTION
+
+### Step 3.1 — Identify Extractable Content
+
+**EXTRACT these types:**
+- Frameworks with named steps (e.g., CLOSER, Farm System)
+- Specific numbers, percentages, benchmarks
+- Process sequences (do X, then Y, then Z)
+- Decision criteria ("hire when...", "fire if...")
+- Compensation structures with figures (OTE, base, commission)
+- Tool/software recommendations with context
+- Organizational structures with ratios
+- Quotes that encapsulate key principles
+
+**SKIP these types:**
+- Filler phrases, introductions, outros
+- Tangential stories without actionable insight
+- Opinions without supporting framework
+- Repeated content already captured
+- Off-topic tangents
+
+### Step 3.2 — Theme Classification
+
+Use this decision tree for EACH insight:
+```
+INSIGHT: [content to classify]
+
+Q1: Is this about PEOPLE structure or roles?
+  → YES: 01-ESTRUTURA-TIME/
+
+Q2: Is this about SELLING process or techniques?
+  → YES: 02-PROCESSO-VENDAS/
+
+Q3: Is this about HIRING or onboarding?
+  → YES: 03-CONTRATACAO/
+
+Q4: Is this about COMPENSATION or commissions?
+  → YES: 04-COMISSIONAMENTO/
+
+Q5: Is this about MEASURING performance?
+  → YES: 05-METRICAS/
+
+Q6: Is this about QUALIFYING leads or pipeline?
+  → YES: 06-FUNIL-APLICACAO/
+
+Q7: Is this about PRICING strategy?
+  → YES: 07-PRICING/
+
+Q8: Is this about TOOLS or tech stack?
+  → YES: 08-FERRAMENTAS/
+
+Q9: Is this about MANAGING people?
+  → YES: 09-GESTAO/
+
+Q10: Is this about CULTURE or motivation?
+  → YES: 10-CULTURA-GAMIFICACAO/
+
+Q11: None of the above?
+  → 99-SECUNDARIO/
+```
+
+### Step 3.3 — Cross-Theme Detection
+
+When insight spans multiple themes:
+```
+EXAMPLE: "Commission structure for closers = 10% base + 5% bonus"
+
+ANALYSIS:
+├── Theme A: 04-COMISSIONAMENTO (compensation details) ← PRIMARY
+└── Theme B: 02-PROCESSO-VENDAS (closer role context)  ← SECONDARY
+
+ACTION:
+1. Write full insight to PRIMARY theme file
+2. Add cross-reference to SECONDARY theme file:
+
+   > **Relacionado:** Estrutura de comissão para closers
+   > → Ver: `/knowledge/SOURCES/{FONTE}/04-COMISSIONAMENTO/closer-compensation.md`
+```
+
+**Primary Theme Selection Rule:**
+Choose the theme where the SPECIFIC DETAIL lives:
+- Numbers/percentages → theme of the metric type
+- Process steps → theme of the process domain
+- Role descriptions → 01-ESTRUTURA-TIME
+- Tool configurations → 08-FERRAMENTAS
+
+---
+
+## PHASE 4: WRITING
+
+### Step 4.1 — File Path Resolution
+```
+TARGET_PATH = /knowledge/SOURCES/[SOURCE-NAME]/[XX-THEME-FOLDER]/[topic-slug].md
+
+Naming convention:
+- [SOURCE-NAME]: Nome da pessoa/empresa (CAIXA ALTA)
+- [XX-THEME-FOLDER]: Código do tema (ex: 04-COMISSIONAMENTO)
+- [topic-slug]: Lowercase, hyphenated
+- Examples:
+  - /knowledge/SOURCES/HORMOZI/04-COMISSIONAMENTO/closer-compensation.md
+  - /knowledge/SOURCES/cole-gordon/01-ESTRUTURA-TIME/farm-system.md
+```
+
+### Step 4.2 — Intelligent Merge Protocol
+```
+IF file EXISTS at TARGET_PATH:
+
+  1. READ existing content
+
+  2. CHECK for source section:
+     IF "## Visão $SOURCE_NAME" section exists:
+       → SCAN for duplicate insights
+       → APPEND only NEW insights to existing section
+       → ENRICH existing points with new specifics (don't duplicate)
+
+     IF section does NOT exist:
+       → APPEND new source section at end of file
+
+IF file does NOT exist:
+  → CREATE with template:
+
+  # [Topic Title]
+
+  ## Visão $SOURCE_NAME ($SOURCE_CODE)
+
+  ### Contexto
+  [Source context: type of content, speaker credentials if mentioned]
+
+  ### [Subtopic]
+  - Key point with specifics
+  - Include numbers: X%, $Y, Z ratio
+
+  ### Fonte
+  - Arquivo: $ARGUMENTS
+  - Processado: [timestamp]
+  - Confiabilidade: Alta
+```
+
+### Step 4.3 — Cross-Reference Injection
+
+For each SECONDARY theme identified in 3.3:
+```
+OPEN secondary theme file (or create if missing)
+LOCATE or CREATE section: ## Referências Cruzadas
+APPEND (if not duplicate):
+
+- **[Insight summary]** → `[relative path to primary file]`
+```
+
+---
+
+## PHASE 5: FINALIZATION
+
+### Step 5.1 — Update Registry
+```
+UPDATE /system/REGISTRY/processed-files.md:
+  ADD entry with:
+  - Hash: [generated code, e.g., SS002]
+  - Arquivo: [filename]
+  - Fonte: $SOURCE_NAME
+  - Data: [timestamp]
+  - Temas: [list of themes updated]
+  - Status: SUCCESS | PARTIAL | FAILED
+```
+
+### Step 5.2 — Update Session State
+```
+UPDATE /system/SESSION-STATE.md:
+  - Add to "Arquivos Processados" table
+  - Update "Knowledge Bases Populadas" if new themes added
+  - Update "Funções Identificadas" with new role counts
+  - Increment version if structural changes made
+```
+
+### Step 5.3 — Generate Execution Report
+
+Display visual timeline during execution:
+```
+┌─ 📥 INPUT
+│  └─ $ARGUMENTS
+│
+├─ 🔍 ANÁLISE ← ATUAL
+│  ├─ Fonte: $SOURCE_NAME
+│  └─ Tipo: $SOURCE_TYPE
+│
+├─ 📚 GLOSSÁRIO
+│  └─ [X] termos verificados/adicionados
+│
+├─ 🧠 EXTRAÇÃO
+│  └─ [X] insights extraídos
+│
+├─ ✏️ ESCRITA
+│  └─ [list of files created/updated]
+│
+├─ 👥 AGENTES
+│  └─ [roles tracked, any CRITICAL flags]
+│
+└─ ✅ REGISTRO
+   └─ Registry e Session State atualizados
+```
+
+Final report:
+```
+═══════════════════════════════════════════
+EXTRACTION COMPLETE: $SOURCE_NAME ($SOURCE_CODE)
+═══════════════════════════════════════════
+Source: $ARGUMENTS
+Status: SUCCESS | PARTIAL | FAILED
+
+Insights Extracted: [count]
+Themes Updated: [list with file paths]
+Cross-References Added: [count]
+Glossary Terms Added: [count]
+Roles Tracked: [list with counts]
+
+New Agents Flagged: [any roles that hit CRITICAL]
+Warnings: [any issues encountered]
+═══════════════════════════════════════════
+```
+
+---
+
+## ERROR RECOVERY PROTOCOLS
+
+| Error | Recovery Action |
+|-------|-----------------|
+| File not found | Log error, exit FAILED |
+| Permission denied | Log warning, skip file, continue |
+| Malformed content | Extract what's parseable, log warning |
+| Theme folder missing | Create folder with UPPERCASE name, continue |
+| Duplicate detection failed | Default to APPEND with source tag |
+| Session state corrupted | Rebuild from processed-files.md |
+| Glossary file missing | Create from template in /system/GLOSSARY/ |
+| Role-tracking missing | Create empty tracking file, continue |
+
+---
+
+## VALIDATION CHECKLIST
+
+Before marking as SUCCESS, verify:
+
+| ✅ | Check | Action if Failed |
+|----|-------|------------------|
+| [ ] | All insights have source attribution | Add $SOURCE_CODE to orphaned insights |
+| [ ] | No duplicate content in theme files | Remove duplicates, keep most detailed |
+| [ ] | Glossary terms use standard names | Replace with glossary canonical form |
+| [ ] | Cross-references are bidirectional | Add missing reverse references |
+| [ ] | Registry entry exists | Create entry in processed-files.md |
+| [ ] | Session state updated | Update SESSION-STATE.md |
+
+---
+
+## DRY-RUN MODE
+
+If `$DRY_RUN == true`:
+```
+→ Show all planned writes WITHOUT executing
+→ Display classification decisions with reasoning
+→ List glossary terms that would be added
+→ Show role counts that would be updated
+→ Useful for testing classification logic before committing
+```
+
+---
+
+## EXECUTION START
+
+```
+Ready to process: $ARGUMENTS
+Beginning PHASE 1: INITIALIZATION...
+```