Vitto: Resumen Completo del Sistema

Documento generado: 2026-03-27 | Audiencia: Desarrolladores del equipo GoGrab

1. Que es Vitto

Vitto es el Super-Agente AI de GoGrab: un sistema autonomo e inteligente que opera a traves de multiples canales (web, desktop, mobile, extensiones, WhatsApp, Slack, Discord, etc.). No es un chatbot — es un agente con conversacion como una de sus muchas modalidades.

Capacidades principales:

• Conversar con memoria persistente y conciencia de contexto
• Ejecutar tareas complejas en fases (ANALYZE -> PLAN -> EXECUTE -> VERIFY)
• Delegar trabajo a sub-agentes coordinados
• Generar modulos, plataformas y codigo completo
• Operar de forma proactiva (notificaciones, heartbeats)
• Controlar maquinas locales (shell, filesystem, screenshots)

2. Arquitectura General

┌──────────────────────────────────────────────────────────────────┐
│                     FRONTEND (Angular 20)                        │
├────────────┬──────────┬───────────┬──────────┬──────────────────┤
│ gograb-    │ gograb-  │ vitto-    │ vitto-   │ vitto-chrome-    │
│ space:6200 │ ai:6400  │ desktop   │ mobile   │ extension        │
│            │          │ (Tauri)   │ (Cap.)   │ (MV3)            │
├────────────┴──────────┴───────────┴──────────┴──────────────────┤
│ vitto-chat │ vitto-hub │ vitto-core │ vitto-maker │ vitto-intel │
├──────────────────────────────────────────────────────────────────┤
│                     BACKEND (NestJS 11)                          │
├──────────┬──────────┬───────────┬───────────────────────────────┤
│ backend- │ backend- │ backend-  │ backend-vitto-api             │
│ api:3000 │ ai:3001  │ docs:3002 │ :3003                         │
├──────────┴──────────┴───────────┴───────────────────────────────┤
│ MongoDB (vitto-mind, vitto-users) │ Redis │ Cloudflare R2       │
└──────────────────────────────────────────────────────────────────┘

3. Apps (Clientes)

3.1 Vitto Desktop (Tauri v2) — apps/vitto-desktop/

• Version: 0.4.0
• Plataformas: macOS, Windows, Linux
• Features: System tray (3 estados), global hotkey (Cmd+Ctrl+V), auto-update via GitHub Releases
• Control local: Shell commands, read/write filesystem, screenshots, AppleScript, clipboard, app detection
• Auth: Browser -> deep link vitto://auth?token=...
• Build: cargo tauri build → .dmg, .msi, .AppImage

3.2 Vitto Mobile (Capacitor 8) — apps/vitto-mobile/

• Version: 0.3.0
• Plataformas: iOS, Android
• Comparte: El mismo frontend Angular (gograb-ai)
• Plugins: SplashScreen, PushNotifications, StatusBar, Keyboard
• Build: npx cap sync → Xcode / Android Studio

3.3 Chrome Extension (MV3) — apps/vitto-chrome-extension/

• Version: 0.2.9
• Features: Side panel, popup, auth sync desde gograb.space, MCP WebSocket, env detection
• Shortcuts: Ctrl+Shift+G (toggle panel), Alt+G (popup)
• Archivos clave: service-worker.js (32KB), mcp-connector.js (60KB), sidepanel.js (23KB)

3.4 VSCode Extension — apps/vitto-vscode-extension/

• Version: 0.15.3
• Framework: TypeScript + Lit 3.x (webview)
• 20+ commands: preflight, logDecision, reviewCode, generateTests, explainCode, findSimilarPatterns
• Providers: CodeLens (decision markers), Code Actions (review/test/explain), Diagnostics
• Webview tabs: Dashboard, Decisions, Intelligence, Agents, Sessions, Repositories

3.5 Local Bridge (Node.js) — apps/vitto-local-bridge/

• Version: 1.0.0
• Proposito: Conecta maquina local al Vitto en la nube (Railway)
• Protocolo: Socket.IO → namespace /local-bridge
• 5 tools: run_command, read_file, write_file, screen_capture, list_directory
• Uso: node index.js --userId=<id> --server=http://localhost:3003

4. Librerias Frontend

4.1 vitto-core (libs/libs-ai/vitto-core/ → @libs-ai/vitto-core)

Servicios fundacionales compartidos por todo el ecosistema.

Modelos: Observatory DTOs, Memory v2, Autonomy level mappers.

4.2 vitto-chat (libs/libs-ai/vitto-chat/ → @libs-ai/vitto-chat)

Motor de chat unificado con 28+ componentes UI.

Componentes clave:

• Chat shell, header, session sidebar, message list
• Input bar con slash commands, suggestion chips, file attachments
• Tool renderers (shell, browser, screen capture, search, delegation)
• Generation panels (image, module, video, widget, prompt)
• Generative UI (VittoRenderComponent, skill preview cards)
• Ambient bar, connectivity panel

4.3 vitto-hub (libs/libs-ai/vitto-hub/ → @libs-ai/vitto-hub)

Dashboard de orquestacion — 44+ componentes.

Servicios: VittoMonitorService (SSE + REST), ObservatoryDataService, VittoConcienceService, VittoIntrospectionService, ComputerUseDataService.

4.4 vitto-intelligence (libs/libs-ai/vitto-intelligence/)

Perfilado de usuarios y mastery tracking.

• API: Dossier CRUD, learn facts, record corrections, update expertise, rebuild brief, delete (GDPR)
• Job Mastery: Playbooks, skills, levels (intern → principal)
• 6 componentes: JobMasteryDashboard, JobMasteryDetail, PlaybookEditor/Creator, UserDossierView, ExecutiveInstructions

4.5 vitto-maker (libs/libs-ai/vitto-maker/)

Constructor interactivo del perfil de Vitto.

• 5 segmentos: capabilities (skills, workflows), knowledge (memories, identity), character (personas, voice), connections (channels, webhooks), governance (trust, autonomy)
• Features: Signal-based state, lazy loading, undo/redo, completeness tracking

4.6 vitto-studio (libs/libs-structural/vitto-studio/)

IDE de plataforma con dock panels.

• 19 dock panels: File tree, code editor, git, deploy, data browser, entity designer, visual canvas, preview, API tester, etc.
• 40+ servicios: VFS, preview, generation, git, deploy
• Estado: Parcial — bugs criticos pendientes (ver seccion 7)

4.7 Librerias de Soporte

5. Backend: backend-vitto-api (Puerto 3003)

5.1 Modulos Principales (24)

5.2 Bases de Datos

5.3 Comunicacion Inter-Servicios

backend-vitto-api:3003
  ├── HTTP → backend-ai-api:3001
  │     ├── POST /ai-api/orchestration/mcp-gateway/execute
  │     ├── GET  /ai-api/infrastructure/decisions
  │     ├── GET  /ai-api/infrastructure/observatory/routines
  │     └── POST /ai-api/sessions/:id/relay
  ├── HTTP → backend-api:3000
  │     └── CRUD operations (platforms, spaces, users)
  ├── HTTP → backend-docs-api:3002
  │     └── Documentation sync
  └── WebSocket → backend-ai-api:3001
        └── HandoffWsClient (agent loop events)

5.4 API Endpoints (53+)

5.5 LLM Routing

6. Sistema de Memoria

┌────────────────── CREACION ──────────────────┐
│ 1. Deteccion explicita (regex) → save        │
│ 2. Extraccion LLM (Claude Haiku async)       │
│ 3. Storage → MongoDB user_agent_memories     │
│ 4. Upsert (mismo userId+key → update)        │
└──────────────────────────────────────────────┘

┌────────────────── RECALL ────────────────────┐
│ 1. Load user profile (users collection)      │
│ 2. Load agent memories (user_agent_memories)  │
│ 3. Vector search + text search combinado      │
│ 4. Inject en prompt del LLM                  │
└──────────────────────────────────────────────┘

Scoped Memory (Session 239+):
  - Per-space, per-user con admission scoring
  - Conversation genomes
  - Multi-tier access control
  - Quarantine/review pipeline

7. Estado Actual: Que Funciona y Que Falta

IMPLEMENTADO Y PRODUCTION-READY

BUGS CRITICOS (Blockers)

DEUDA TECNICA

FEATURES INCOMPLETOS

FUNCIONALIDAD FUTURA (Specs Existentes No Implementados)

8. Como Usar Vitto (Desarrollo Local)

8.1 Levantar el Backend

# Terminal 1: Backend principal (auth, users, CRUD)
yarn serve:backend-api          # Puerto 3000

# Terminal 2: Backend AI (chat, agents, MCP)
yarn serve:backend-ai           # Puerto 3001

# Terminal 3: Backend Vitto (super-agent)
yarn serve:backend-vitto        # Puerto 3003

# O todos juntos:
yarn serve:all-backends

8.2 Levantar el Frontend

# App principal con Vitto integrado
yarn serve:gograb-space          # Puerto 6200

# O la app de AI directamente
yarn serve:gograb-ai             # Puerto 6400

8.3 Apps Nativas

# Desktop (Tauri)
cd apps/vitto-desktop && cargo tauri dev

# Mobile (Capacitor)
cd apps/vitto-mobile
npx cap sync
npx cap run ios --livereload --external

# Chrome Extension
# Cargar apps/vitto-chrome-extension/ como extension sin empaquetar

# VSCode Extension
cd apps/vitto-vscode-extension
npm run compile && npm run install:vscode

8.4 Local Bridge (Control Remoto)

# Conectar tu maquina al Vitto cloud
node apps/vitto-local-bridge/index.js \
  --userId=<tu-user-id> \
  --server=http://localhost:3003

8.5 Variables de Entorno

Archivo principal: .env.local (en la raiz del proyecto, git-ignored) Template de referencia: deployment/shared/.env.example Load order: .env.local > .env > process env

Minimo requerido para desarrollo local

Todas las variables por categoria

Secrets & API Keys:

Bases de Datos:

Puertos (NUNCA hardcodear, siempre desde .env.local):

URLs Inter-Servicio:

Config Vitto Super-Agent:

Workspace:

Archivos de configuracion:

9. Grafo de Dependencias

vitto-core (fundacion)
├── VittoApiService → backend-vitto-api:3003
├── VittoScreenCaptureService
├── VittoPageContextRegistry
├── BrainContextService
└── Models: Observatory DTOs, Memory v2

vitto-chat (conversacion)
├── depends: vitto-core, @libs-foundational/api-config
├── VittoChatEngineService → /ai-api/chat (backend-ai:3001)
├── VittoChatPersistenceService → IndexedDB
├── VittoEmotionService
├── VittoVoiceService → Web Speech API
└── 28+ UI components

vitto-hub (dashboard)
├── depends: vitto-core
├── VittoMonitorService → SSE + REST
└── 44+ components (lazy-loaded routes)

vitto-intelligence (perfilado)
├── depends: backend-vitto-api
├── Dossier API + Job Mastery
└── 6 UI components

vitto-maker (configuracion)
├── depends: vitto-core
├── VittoProfileService (5 segmentos)
└── Undo/redo, completeness tracking

vitto-studio (IDE)
├── depends: vitto-core, vitto-chat
├── 19 dock panels, 40+ services
└── VFS, preview, generation, git, deploy

Soporte:
├── agent-mind → Visualizacion 3D
├── agent-orchestrator → Task orchestration
├── agent-sdk → Universal agent framework
├── chat-plugin-core → Plugin system
├── chat-sync → MongoDB persistence
├── prompt-system → Templates
├── platform-intelligence → Knowledge feeding
└── vitto-rive-avatar → Animated avatar (14 moods)

10. Estimacion de Esfuerzo para GA

Total estimado para GA (P0+P1): ~4-6 semanas Full feature set (P0-P3): ~3-4 meses