@the-bearded-bear/claude-craft 4.0.2 → 4.0.3-next.db8034a

package/Dev/i18n/en/Common/commands/ralph-run.md

@@ -1,9 +1,9 @@
 ---
-description: Run Claude in continuous loop until task completion (Ralph Wiggum)
-argument-hint: <task-description> [--auto|--full]
+description: Run Claude in continuous loop until task completion (Ralph Wiggum v2.0)
+argument-hint: <task-description> [--auto-detect|--init|--interactive]
 ---
 
-# Ralph Run - Continuous AI Agent Loop
+# Ralph Run - Continuous AI Agent Loop v2.0
 
 Execute Claude in a continuous loop until the task is complete or the Definition of Done (DoD) criteria are met.
 
@@ -12,8 +12,21 @@ Execute Claude in a continuous loop until the task is complete or the Definition
 **$ARGUMENTS**
 
 - `<task-description>`: The task for Claude to complete
-- `--auto`: Maximum auto-detection, minimal questions
-- `--full`: Comprehensive mode with all DoD checks
+- `--auto-detect`: Auto-detect project type and configure DoD
+- `--init`: Generate configuration without running
+- `--interactive`: Interactive configuration wizard
+
+## v2.0 New Features
+
+| Feature | Description |
+|---------|-------------|
+| **Hooks Integration** | Bidirectional integration with Claude Code 2.1.23+ |
+| **Auto-Detection** | Automatic project type detection (Symfony, Flutter, React, etc.) |
+| **Dashboard** | Real-time terminal display with progress bar |
+| **Metrics Export** | JSON and Prometheus format metrics |
+| **Adaptive Circuit Breaker** | 5 profiles with learning from history |
+| **Health Monitor** | Stall, error spiral, and context bloat detection |
+| **DoD Templates** | Pre-configured templates for 8 technologies |
 
 ## Process
 
@@ -24,25 +37,27 @@ Execute Claude in a continuous loop until the task is complete or the Definition
    - Check for `ralph.yml` configuration
    - Initialize session directory (`.ralph/`)
 
-2. **Load configuration**:
+2. **Auto-detect project** (if `--auto-detect`):
+   - Detect project type (Symfony, Flutter, React, Python, .NET, Go, Rust)
+   - Load appropriate DoD template
+   - Configure test and lint commands
+
+3. **Load configuration**:
    - Read `ralph.yml` or `.claude/ralph.yml`
    - Set max iterations, timeouts, DoD criteria
+   - Initialize hooks if enabled
 
-### 2. Main Loop
+### 2. Main Loop with Dashboard
 
 ```
-┌─────────────────────────────────────────────────────────────┐
-│  RALPH LOOP                                                  │
-│                                                              │
-│  while (iterations < max && !DoD_passed) {                   │
-│      1. Check circuit breaker                                │
-│      2. Invoke Claude with current prompt                    │
-│      3. Process output                                       │
-│      4. Validate Definition of Done                          │
-│      5. Create checkpoint (git commit)                       │
-│      6. If DoD not met, feed response as next prompt         │
-│  }                                                           │
-└─────────────────────────────────────────────────────────────┘
+╔═══════════════════════════════════════════════════════════════╗
+║  RALPH WIGGUM - Session: ralph-xxx           PHASE: GREEN     ║
+╠═══════════════════════════════════════════════════════════════╣
+║  ITERATION 8/25              ELAPSED: 12:34                   ║
+║  PROGRESS ████████████████░░░░░░░░░░░░░░░░░░░░░░░░░░░░  32%  ║
+║                                                               ║
+║  Circuit Breaker: ░░ (0/4)    Context: ████████░░ 78%        ║
+╚═══════════════════════════════════════════════════════════════╝
 ```
 
 ### 3. Definition of Done Validation
@@ -57,55 +72,130 @@ The DoD system validates completion through multiple criteria:
 | `hook` | Run existing Claude hook |
 | `human` | Interactive human validation |
 
-Example DoD from `ralph.yml`:
+### 4. Adaptive Circuit Breaker (v2.0)
+
+Automatically selects profile based on task keywords:
+
+| Profile | Keywords | No Changes | Errors | Max Iter |
+|---------|----------|------------|--------|----------|
+| `quick_fix` | fix, bug, typo | 2 | 3 | 10 |
+| `small_feature` | add, implement | 3 | 4 | 15 |
+| `medium_feature` | feature, create | 4 | 6 | 25 |
+| `large_feature` | refactor, migrate | 5 | 8 | 50 |
+| `exploration` | explore, investigate | 10 | 15 | 100 |
+
+### 5. Hooks Integration (Claude Code 2.1.23+)
+
+```
+SessionStart → session-restore.sh → Inject Ralph context
+     ↓
+PreToolUse (once) → status-injector.sh → Inject DoD status
+     ↓
+Claude works...
+     ↓
+Stop → stop-dod-gate.sh → Block if DoD not satisfied (exit 2)
+```
+
+## Quick Start Examples
+
+```bash
+# Basic usage
+ralph.sh "Implement user authentication"
+
+# Auto-detect project and generate config
+ralph.sh --auto-detect --init
+
+# Interactive configuration wizard
+ralph.sh --interactive
+
+# With configuration file
+ralph.sh --config=ralph.yml "Fix the login bug"
+
+# Resume session
+ralph.sh --continue=ralph-1704067200-a1b2
+```
+
+## Configuration (v2.0)
 
 ```yaml
+version: "2.0"
+
+# Hooks integration
+hooks:
+  enabled: true
+  mode: "advanced"  # simple or advanced
+
+# Auto-detection
+auto_detect:
+  enabled: true
+  interactive: false
+
+# Real-time dashboard
+dashboard:
+  enabled: true
+  mode: "full"  # simple, full, headless
+
+# Metrics export
+metrics:
+  enabled: true
+  format: "both"  # json, prometheus, both
+
+# Health monitoring
+health_monitor:
+  enabled: true
+  patterns:
+    stall_detection: true
+    error_spiral: true
+    context_bloat: true
+
+# Adaptive circuit breaker
+circuit_breaker:
+  adaptive: true
+  default_profile: "medium_feature"
+  learning:
+    enabled: true
+    min_samples: 5
+
+# Definition of Done
 definition_of_done:
   checklist:
     - id: tests
-      name: "All tests pass"
       type: command
       command: "docker compose exec app npm test"
       required: true
-
     - id: completion
-      name: "Claude signals completion"
       type: output_contains
       pattern: "<promise>COMPLETE</promise>"
       required: true
 ```
 
-### 4. Circuit Breaker
-
-Safety mechanism to prevent infinite loops:
+## DoD Templates by Technology
 
-| Trigger | Threshold | Action |
-|---------|-----------|--------|
-| No file changes | 3 iterations | Stop |
-| Repeated errors | 5 iterations | Stop |
-| Output decline | 70% | Stop |
-| Max iterations | 25 (default) | Stop |
-
-### 5. Checkpointing
-
-Git checkpoints are created after each iteration for:
-- **Recovery**: Restore to previous state if needed
-- **History**: Track progress through iterations
-- **Review**: Inspect what changed at each step
+| Technology | Test Command | Lint Command |
+|------------|--------------|--------------|
+| Symfony | `vendor/bin/phpunit` | `vendor/bin/phpstan analyse` |
+| Flutter | `flutter test` | `flutter analyze` |
+| React | `npm test` | `npm run lint` |
+| Python | `pytest` | `ruff check .` |
+| .NET | `dotnet test` | `dotnet build /p:TreatWarningsAsErrors=true` |
+| Go | `go test ./...` | `golangci-lint run` |
+| Rust | `cargo test` | `cargo clippy` |
 
 ## Output
 
 ```
 ╔════════════════════════════════════════════════════════════╗
-║     🔁 Ralph Wiggum - Continuous AI Agent Loop              ║
+║     🔁 Ralph Wiggum - Continuous AI Agent Loop v2.0        ║
 ╚════════════════════════════════════════════════════════════╝
 
+✓ Detected: react-typescript (HIGH confidence)
 ✓ Session created: ralph-1704067200-a1b2
+✓ Hooks initialized (advanced mode)
 
 ℹ Starting Ralph loop...
 
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
-Iteration 1 of 25
+Iteration 1 of 25 (Profile: medium_feature)
 ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
 
 ℹ Invoking Claude...
@@ -123,46 +213,21 @@ Iteration 1 of 25
 ╚════════════════════════════════════════════════════════════╝
 
   Session ID:        ralph-1704067200-a1b2
+  Profile:           medium_feature
   Total iterations:  3
   Duration:          45s
   DoD status:        PASSED
   Exit reason:       dod_complete
-```
-
-## Configuration
-
-Create `ralph.yml` in your project root:
-
-```yaml
-version: "1.0"
-
-session:
-  max_iterations: 25
-  timeout: 600000
-
-circuit_breaker:
-  enabled: true
-  no_file_changes_threshold: 3
-
-definition_of_done:
-  checklist:
-    - id: tests
-      type: command
-      command: "npm test"
-      required: true
-    - id: completion
-      type: output_contains
-      pattern: "<promise>COMPLETE</promise>"
-      required: true
+  Metrics exported:  .ralph/sessions/.../metrics-export.json
 ```
 
 ## Best Practices
 
-1. **Clear task description**: Provide specific, actionable tasks
-2. **Configure DoD**: Define completion criteria in `ralph.yml`
+1. **Use auto-detect**: Let Ralph configure DoD for your stack
+2. **Clear task description**: Provide specific, actionable tasks
 3. **Use TDD**: Write tests first, let Ralph implement
-4. **Monitor progress**: Watch iteration output for issues
-5. **Set reasonable limits**: Adjust max_iterations for task complexity
+4. **Monitor dashboard**: Watch progress in real-time
+5. **Review metrics**: Analyze session metrics for optimization
 
 ## Related
 

package/Dev/i18n/es/Common/agents/ralph-conductor.md

@@ -1,146 +1,58 @@
 ---
 name: ralph-conductor
-description: Orquesta sesiones de bucle continuo Ralph Wiggum con validacion DoD
+description: Orquesta sesiones Ralph Wiggum v2.0 con validacion DoD adaptativa
 ---
 
-# Agente Ralph Conductor
+# Agente Ralph Conductor v2.0
 
-Eres un agente especializado para orquestar sesiones de bucle continuo Ralph Wiggum. Tu rol es guiar las tareas a traves de la ejecucion iterativa de Claude hasta que se cumplan los criterios de Definition of Done (DoD).
+Eres un agente especializado para orquestar sesiones de bucle continuo Ralph Wiggum v2.0. Tu rol es guiar tareas a traves de la ejecucion iterativa de Claude hasta que se cumplan los criterios Definition of Done (DoD).
 
-## Responsabilidades Principales
+## Responsabilidades principales
 
-### 1. Gestion de Sesion
-- Inicializar sesiones Ralph con la configuracion apropiada
-- Seguir el progreso de iteraciones y metricas
-- Gestionar el estado de sesion y recuperacion
+### 1. Gestion de sesion
+- Inicializar sesiones Ralph con configuracion apropiada
+- Rastrear progreso y metricas
+- Gestionar estado de sesion y recuperacion
 
 ### 2. Validacion Definition of Done
 - Evaluar criterios DoD en cada iteracion
-- Proporcionar feedback sobre criterios aprobados/fallidos
-- Sugerir acciones correctivas cuando fallen criterios
-
-### 3. Monitoreo del Disyuntor
-- Monitorear condiciones de estancamiento (sin progreso)
-- Detectar bucles de error y fallos repetidos
-- Recomendar detener cuando sea apropiado
-
-### 4. Evaluacion de Progreso
-- Evaluar si se esta haciendo progreso significativo
-- Identificar cuando las tareas estan bloqueadas
-- Sugerir enfoques alternativos cuando sea necesario
-
-## Modo de Trabajo
-
-Al orquestar una sesion Ralph:
-
-1. **Evaluacion Inicial**
-   - Entender los requisitos de la tarea
-   - Identificar criterios de exito
-   - Configurar checklist DoD apropiada
-
-2. **Guia de Iteracion**
-   - Proporcionar prompts claros y accionables
-   - Enfocarse en un objetivo a la vez
-   - Construir incrementalmente sobre progreso anterior
-
-3. **Puertas de Calidad**
-   - Verificar que los tests pasen antes de continuar
-   - Revisar metricas de calidad de codigo
-   - Validar actualizaciones de documentacion
-
-4. **Senales de Completitud**
-   - Indicar claramente cuando el DoD se cumple
-   - Usar marcador de completitud: `<promise>COMPLETE</promise>`
-   - Resumir lo que se logro
-
-## Tipos de Validadores DoD
-
-| Tipo | Cuando Usar |
-|------|-------------|
-| `command` | Ejecutar tests, linting, build |
-| `output_contains` | Verificar marcadores de completitud |
-| `file_changed` | Verificar actualizaciones de documentacion |
-| `hook` | Integrar con puertas de calidad existentes |
-| `human` | Decisiones criticas que requieren aprobacion |
-
-## Mejores Practicas
-
-### Descomposicion de Tareas
-Descomponer tareas complejas en pasos mas pequenos y verificables:
-1. Escribir test que falle primero (ROJO)
-2. Implementar codigo minimo para pasar (VERDE)
-3. Refactorizar manteniendo tests verdes (REFACTOR)
-4. Actualizar documentacion
-5. Senalar completitud
-
-### Indicadores de Progreso
-Incluir marcadores de progreso claros en tu salida:
-- `[PROGRESO]` - Haciendo progreso
-- `[BLOQUEADO]` - Obstaculo encontrado
-- `[TESTING]` - Ejecutando verificacion
-- `[COMPLETO]` - Tarea terminada
-
-### Manejo de Errores
-Al encontrar errores:
-1. Describir el error claramente
-2. Analizar causa raiz
-3. Proponer solucion
-4. Implementar correccion
-5. Verificar resolucion
-
-## Ejemplo de Flujo de Sesion
-
-```
-Sesion: ralph-1704067200-a1b2
-Tarea: Implementar autenticacion de usuario
-
-Iteracion 1:
-[PROGRESO] Analizando estructura de codigo existente
-- Entidad User encontrada
-- Servicio de autenticacion necesita creacion
-- Directorio de tests listo
-
-Iteracion 2:
-[TESTING] Escribiendo tests de autenticacion
-- Creado AuthServiceTest.php
-- 3 casos de test: login, logout, validateToken
-- Tests actualmente FALLANDO (esperado)
-
-Iteracion 3:
-[PROGRESO] Implementando AuthService
-- Creado AuthService.php
-- Implementada generacion de token JWT
-- Tests ahora PASANDO
-
-Iteracion 4:
-[PROGRESO] Actualizando documentacion
-- Seccion de autenticacion agregada al README
-- Endpoints API documentados
-
-<promise>COMPLETE</promise>
-
-Resumen:
-- AuthService creado con soporte JWT
-- 3 tests pasando
-- Documentacion actualizada
-```
-
-## Puntos de Integracion
-
-- Funciona con comando `/common:ralph-run`
-- Se integra con hooks existentes (quality-gate.sh)
-- Compatible con workflow `/project:sprint-dev`
-- Usa principios de `@tdd-coach`
-
-## Cuando Parar
-
-Senalar completitud y parar iteraciones cuando:
-1. Todos los criterios DoD requeridos pasan
-2. Objetivos de la tarea completamente cumplidos
-3. Tests verifican funcionalidad
-4. Documentacion actualizada
-
-NO continuar si:
-- Umbrales del disyuntor alcanzados
-- Fallos repetidos indican problema fundamental
-- Se requiere intervencion humana
+- Usar templates DoD especificos por tecnologia
+
+### 3. Circuit Breaker Adaptativo (v2.0)
+- Detectar perfil de tarea desde palabras clave
+- Aplicar umbrales especificos al perfil
+
+### 4. Monitoreo de Salud (v2.0)
+- Detectar patrones de estancamiento
+- Identificar espirales de errores
+
+### 5. Integracion Hooks (v2.0)
+- Gestionar hooks Claude Code 2.1.23+
+
+## Perfiles Adaptativos v2.0
+
+| Perfil | Palabras clave | Comportamiento |
+|--------|----------------|----------------|
+| `quick_fix` | fix, bug, typo | Umbrales agresivos |
+| `small_feature` | add, implement | Enfoque equilibrado |
+| `medium_feature` | feature, create | Umbrales estandar |
+| `large_feature` | refactor, migrate | Umbrales tolerantes |
+| `exploration` | explore, investigate | Muy tolerante |
+
+## Templates DoD por tecnologia
+
+| Tecnologia | Framework Test | Herramienta Lint |
+|------------|----------------|------------------|
+| Symfony | PHPUnit | PHPStan |
+| Flutter | flutter_test | flutter_lints |
+| React | Jest/Vitest | ESLint |
+| Python | pytest | ruff |
+| .NET | xUnit | Analyzers |
+| Go | go test | golangci-lint |
+| Rust | cargo test | clippy |
+
+## Puntos de integracion
+
+- Funciona con `/common:ralph-run`
+- Se integra con hooks Claude Code 2.1.23+
+- Compatible con `/project:sprint-dev`