@the-bearded-bear/claude-craft 8.23.0 → 8.23.1-next.6e7d46b

package/Dev/i18n/base/Common/templates/AGENTS.md.template

@@ -0,0 +1,92 @@
+# Agents disponibles pour ce projet
+
+> Ce fichier liste les agents Claude Code installes via claude-craft.
+> Format compatible avec le standard emergent AGENTS.md (https://agents.md).
+
+## Comment utiliser
+
+- Invoque un agent dans Claude Code : `@nom-agent <ta demande>`
+- Liste complete : voir `.claude/agents/`
+- Documentation : voir `docs/AGENTS.md`
+
+## Agents par categorie
+
+### Architecture et design
+
+- `@api-designer` — Conception API REST/GraphQL, OpenAPI 3.1.1, GraphQL Federation
+- `@database-architect` — Conception et optimisation BDD (PostgreSQL, MySQL, schemas)
+- `@refactoring-specialist` — Refactoring sur, patterns SOLID, extraction methode
+- `@ui-designer` — Design systems, Tailwind, design tokens
+- `@ux-ergonome` — Ergonomie cognitive, user flows, heuristiques
+
+### Code review (par stack)
+
+- `@symfony-reviewer` — Symfony 8.1 / PHP 8.4+ (8.5 recommandé), DDD, Doctrine, API Platform
+- `@laravel-reviewer` — Laravel 13 / PHP 8.5, Pest 4/5, Eloquent, Sanctum
+- `@react-reviewer` — React 19.3 + Compiler 1.0, TypeScript, hooks, bundle analysis
+- `@vuejs-reviewer` — Vue 3.5+, Composition API, Pinia 4, Vitest 5
+- `@angular-reviewer` — Angular 22, Signals, Standalone, Zoneless par défaut
+- `@flutter-reviewer` — Flutter 3.47 / Dart 3.13, BLoC v9, Riverpod 3
+- `@reactnative-reviewer` — React Native 0.87 New Architecture (JSI, TurboModules)
+- `@python-reviewer` — Python 3.15 / 3.14+, FastAPI, Pydantic v2, async correctness
+- `@csharp-reviewer` — C# 14 / .NET 10, Clean Architecture, CQRS, EF Core
+- `@php-reviewer` — PHP 8.5, Property Hooks, PER-CS 3.0, PHPStan level max, Pest 5
+- `@paperclip-reviewer` — Paperclip 2026.x, governance-first, Node.js + PostgreSQL
+- `@vite-reviewer` — Vite 8.x framework-agnostic : apps vanilla JS/TS, `build.lib`, multi-pages, Workers/WASM
+- `@vercel-reviewer` — Vercel : `vercel.json`, Functions (Node.js/Fluid Compute), ISR, Cron Jobs, Storage
+
+### Test et qualite
+
+- `@tdd-coach` — Coach TDD/BDD (red/green/refactor), Gherkin
+- `@performance-auditor` — Profilage, optimisation, bundle size, Core Web Vitals
+
+### Securite
+
+- `@security-auditor` — Audit OWASP Top 10:2025, SAST, secrets detection, authZ/N
+
+### Operations
+
+- `@devops-engineer` — Docker, K8s, OpenTofu, Ansible, Coolify, FrankenPHP
+- `@observability-engineer` — OpenTelemetry, traces, structured logs, metrics
+- `@chaos-engineer` — Resilience testing, Litmus, Gremlin, fault injection
+- `@migration-specialist` — Schema migrations zero-downtime, version upgrades
+- `@cost-optimizer` — FinOps, right-sizing, caching, LLM token reduction
+
+### Donnees et ML
+
+- `@data-analyst` — SQL optimization, metrics design, BI dashboards
+- `@mlops-engineer` — ML pipelines, MLflow, Kubeflow, feature stores
+
+### Accessibilite
+
+- `@accessibility-expert` — Audit WCAG 2.2 AAA, ARIA, keyboard navigation
+
+### Developer experience
+
+- `@devex-engineer` — CLI design, onboarding, tooling, DX metrics
+- `@research-assistant` — Recherche technique, documentation, veille
+
+### Orchestration
+
+- `@uiux-orchestrator` — Coordination UI/UX (designer + ergonome + a11y)
+- `@ralph-conductor` — Sessions Ralph Wiggum continues avec DoD adaptative
+
+## Frontmatter recommande
+
+Pour personnaliser un agent local, override via frontmatter dans `.claude/agents/<nom>.md` :
+
+```yaml
+---
+model: claude-haiku-5-5  # economiser tokens pour reviewers read-only
+effort: low                  # ou medium / high
+maxTurns: 20                 # limite anti-runaway
+disallowedTools: [Bash]      # securite : pas d execution shell
+context: fork                # isoler le contexte (skills lourds)
+---
+```
+
+## En savoir plus
+
+- Documentation complete : [docs/AGENTS.md](docs/AGENTS.md)
+- Standard AGENTS.md : https://agents.md
+- claude-craft : https://github.com/TheBeardedBearSAS/claude-craft

package/Dev/i18n/de/Common/rules/12-context-management.md

@@ -492,12 +492,14 @@ Bei der Kompaktierung immer erhalten:
 
 | Variable | Beschreibung |
 |----------|-------------|
-| `CLAUDE_CODE_SUBAGENT_MODEL` | **Standard**-Modell für nicht typisierte Sub-Agents (z. B. `sonnet`). **Das `model:`-Frontmatter eines Agents hat Vorrang**: 11 Reviewer mit `model: haiku` bleiben auf Haiku, auch wenn diese Variable auf `sonnet` gesetzt ist. |
+| `CLAUDE_CODE_SUBAGENT_MODEL` | **Standard**-Modell für nicht typisierte Sub-Agents (z. B. `sonnet`). **Seit Claude Code 2.1.251** hat das `model:`-Frontmatter eines Agents Vorrang: die 11 Reviewer mit `model: haiku` bleiben auf Haiku. **Vor 2.1.251 war es umgekehrt** — diese Variable überschrieb das Frontmatter. |
 | `CLAUDE_CODE_FORK_SUBAGENT` | `1`, um den Kontext von Sub-Agents zu isolieren (Forked Subagents, v2.1.117+) |
 | `ENABLE_PROMPT_CACHING_1H` / `FORCE_PROMPT_CACHING_5M` | Prompt-Caching verlängern/erzwingen (−40 % bei wiederholenden Sitzungen) |
 | `CLAUDE_CODE_DISABLE_AUTO_MEMORY` | Auf `1` setzen, um automatischen Speicher zu deaktivieren |
 
-> **Modell-Vorrang (Audit 2026-06-08):** `model:` im Agent-Frontmatter > `CLAUDE_CODE_SUBAGENT_MODEL` > Sitzungsmodell. Die Variable `SUBAGENT_MODEL=sonnet` überschreibt NICHT explizit auf `model: haiku` gesetzte Agents — sie gilt nur für Sub-Agents ohne definiertes `model:`-Feld.
+> **Modell-Vorrang — gültig seit Claude Code 2.1.251:** `model:` aus dem Aufruf > `model:` im Agent-Frontmatter (wobei `inherit` das Modell der Hauptkonversation übernimmt) > `CLAUDE_CODE_SUBAGENT_MODEL` > Sitzungsmodell.
+>
+> **⚠️ Vor 2.1.251 war die Reihenfolge umgekehrt:** `CLAUDE_CODE_SUBAGENT_MODEL` stand an erster Stelle und überschrieb sowohl den Aufrufparameter als auch das Frontmatter, einschließlich `model: inherit`. Auf diesen Versionen liefen die `model: haiku`-Reviewer mit `SUBAGENT_MODEL=sonnet` tatsächlich auf Sonnet — das Gegenteil der Absicht. Prüfe `claude --version`, bevor du dich auf diese Tabelle verlässt. Quelle: [Sub-Agents-Doku](https://code.claude.com/docs/en/sub-agents).
 
 ---
 

package/Dev/i18n/de/Common/templates/settings.json.template

@@ -18,19 +18,10 @@
       "Write",
       "Glob",
       "Grep",
+      "__STACK_PERMISSIONS__",
       "Bash(git:*)",
-      "Bash(npm run:*)",
-      "Bash(npm test:*)",
-      "Bash(pnpm:*)",
-      "Bash(yarn:*)",
-      "Bash(composer:*)",
-      "Bash(php bin/console:*)",
-      "Bash(php artisan:*)",
-      "Bash(dotnet:*)",
       "Bash(make:*)",
-      "Bash(docker compose:*)",
-      "Bash(flutter:*)",
-      "Bash(ng:*)"
+      "Bash(docker compose:*)"
     ],
     "deny": [
       "Bash(rm -rf:*)",

package/Dev/i18n/de/Workflow/commands/auto-sprint.md

@@ -40,7 +40,7 @@ $ARGUMENTS
 - `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` gesetzt
 - `gh` CLI authentifiziert (PR erstellen / Prüfungen / Merge)
 - Docker verfügbar (alle Tests laufen über Docker — siehe Projekt-CLAUDE.md)
-- BMAD v6-Projekt mit vorhandener `.bmad/sprint-status.yaml`
+- BMAD v6-Projekt: `.bmad/sprint-status.yaml` wird bei Bedarf von `/workflow:start` (Phase 1) angelegt
 
 > Falls eine Voraussetzung fehlt, sofort mit einer klaren, handlungsorientierten Meldung abbrechen. Eine Phase nicht stillschweigend überspringen.
 

package/Dev/i18n/de/Workflow/commands/start.md

@@ -224,7 +224,43 @@ Punkte |
 {Sprint Planning Notizen, getroffene Entscheidungen...}
 ```
 
-### Schritt 6: Finale Checkliste
+### Schritt 6: `.bmad/sprint-status.yaml` synchronisieren
+
+`/gate:validate-sprint` und die `/sprint:*`-Befehle lesen den Sprint aus `.bmad/sprint-status.yaml`. Die Datei jetzt aus der obigen `sprint-goal.md` und den ausgewählten US-Dateien (`project-management/backlog/user-stories/US-*.md`) schreiben:
+
+1. `.bmad/` anlegen, falls nicht vorhanden.
+2. `metadata`: `sprint_id` (`sprint-{N}`), `name`, `start_date`, `end_date`, `goal`, `capacity_points`, aus `sprint-goal.md` übernommen.
+3. `stories`: ein Eintrag pro ausgewählter US: `title`, `status: ready-for-dev`, `story_points`, `epic_id`, `priority` (MoSCoW), `dependencies`, `tasks`, `history: []`.
+4. Datei für **dieselbe** `sprint_id` bereits vorhanden: bestehende Status beibehalten (eine Story nie zurücksetzen) und nur neue US ergänzen.
+5. Datei für einen **anderen** Sprint vorhanden: zuerst deren Stories nach `archived_sprints.<alte sprint_id>` (`status`, `points`) verschieben, dann den neuen Sprint schreiben.
+
+```yaml
+# .bmad/sprint-status.yaml
+version: "1.0"
+metadata:
+  sprint_id: "sprint-2"
+  name: "Benutzerverwaltung"
+  start_date: "2026-11-02"
+  end_date: "2026-11-13"
+  goal: "Benutzer können sich registrieren, anmelden und ihr Profil bearbeiten"
+  capacity_points: 21
+stories:
+  US-010:
+    title: "Benutzerregistrierung"
+    status: "ready-for-dev"
+    story_points: 5
+    epic_id: "EPIC-002"
+    priority: "must"
+    dependencies: ["US-001"]
+    tasks: { total: 0, completed: 0, list: [] }
+    history: []
+archived_sprints:
+  sprint-1:
+    stories:
+      US-001: { status: "done", points: 5 }
+```
+
+### Schritt 7: Finale Checkliste
 
 ```
 ══════════════════════════════════════════════════════════════
@@ -291,6 +327,9 @@ Punkte |
 ║                   NÄCHSTER SCHRITT                        ║
 ╠══════════════════════════════════════════════════════════╣
 ║                                                          ║
+║  → /gate:validate-sprint {N}                             ║
+║    Prüfen, ob der Sprint bereit ist (Gate)               ║
+║                                                          ║
 ║  → /sprint:next-story                                    ║
 ║    Erste Story aus dem Sprint-Backlog auswählen          ║
 ║                                                          ║

package/Dev/i18n/en/Common/rules/12-context-management.md

@@ -492,12 +492,14 @@ During compaction, always preserve:
 
 | Variable | Description |
 |----------|-------------|
-| `CLAUDE_CODE_SUBAGENT_MODEL` | **Default** model for untyped sub-agents (e.g., `sonnet`). **Agent frontmatter `model:` takes precedence**: 11 reviewers with `model: haiku` stay on Haiku even with this variable set to `sonnet`. |
+| `CLAUDE_CODE_SUBAGENT_MODEL` | **Default** model for untyped sub-agents (e.g., `sonnet`). **Since Claude Code 2.1.251**, agent frontmatter `model:` takes precedence: the 11 reviewers with `model: haiku` stay on Haiku. **Before 2.1.251 it was the other way round** — this variable overrode the frontmatter. |
 | `CLAUDE_CODE_FORK_SUBAGENT` | `1` to isolate sub-agent contexts (forked subagents, v2.1.117+) |
 | `ENABLE_PROMPT_CACHING_1H` / `FORCE_PROMPT_CACHING_5M` | Extend/force prompt caching (−40% on repetitive sessions) |
 | `CLAUDE_CODE_DISABLE_AUTO_MEMORY` | Set to `1` to disable automatic memory |
 
-> **Model precedence (audit 2026-06-08):** `model:` in agent frontmatter > `CLAUDE_CODE_SUBAGENT_MODEL` > session model. Setting `SUBAGENT_MODEL=sonnet` does NOT override agents explicitly set to `model: haiku` — it only applies to sub-agents without a defined `model:` field.
+> **Model precedence — valid since Claude Code 2.1.251:** per-invocation `model:` > agent frontmatter `model:` (where `inherit` picks the main conversation's model) > `CLAUDE_CODE_SUBAGENT_MODEL` > session model.
+>
+> **⚠️ Before 2.1.251 the order was reversed:** `CLAUDE_CODE_SUBAGENT_MODEL` came first and overrode both the per-invocation parameter and the frontmatter, including `model: inherit`. On those versions `SUBAGENT_MODEL=sonnet` really did run the `model: haiku` reviewers on Sonnet — the opposite of the intent. Check `claude --version` before relying on this table. Source: [sub-agents docs](https://code.claude.com/docs/en/sub-agents).
 
 ---
 

package/Dev/i18n/en/Common/templates/settings.json.template

@@ -18,19 +18,10 @@
       "Write",
       "Glob",
       "Grep",
+      "__STACK_PERMISSIONS__",
       "Bash(git:*)",
-      "Bash(npm run:*)",
-      "Bash(npm test:*)",
-      "Bash(pnpm:*)",
-      "Bash(yarn:*)",
-      "Bash(composer:*)",
-      "Bash(php bin/console:*)",
-      "Bash(php artisan:*)",
-      "Bash(dotnet:*)",
       "Bash(make:*)",
-      "Bash(docker compose:*)",
-      "Bash(flutter:*)",
-      "Bash(ng:*)"
+      "Bash(docker compose:*)"
     ],
     "deny": [
       "Bash(rm -rf:*)",

package/Dev/i18n/en/Workflow/commands/auto-sprint.md

@@ -40,7 +40,7 @@ $ARGUMENTS
 - `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` set
 - `gh` CLI authenticated (PR create / checks / merge)
 - Docker available (all tests run via Docker — see project CLAUDE.md)
-- BMAD v6 project with `.bmad/sprint-status.yaml` present
+- BMAD v6 project: `.bmad/sprint-status.yaml` is created by `/workflow:start` (phase 1) if missing
 
 > If a prerequisite is missing, abort early with a clear, actionable message. Do not silently skip a phase.
 

package/Dev/i18n/en/Workflow/commands/start.md

@@ -224,7 +224,43 @@ Points |
 {Sprint planning notes, decisions made...}
 ```
 
-### Step 6: Final Checklist
+### Step 6: Sync `.bmad/sprint-status.yaml`
+
+`/gate:validate-sprint` and the `/sprint:*` commands read the sprint from `.bmad/sprint-status.yaml`. Write it now from the `sprint-goal.md` above and the selected US files (`project-management/backlog/user-stories/US-*.md`):
+
+1. Create `.bmad/` if missing.
+2. `metadata`: `sprint_id` (`sprint-{N}`), `name`, `start_date`, `end_date`, `goal`, `capacity_points`, taken from `sprint-goal.md`.
+3. `stories`: one entry per selected US: `title`, `status: ready-for-dev`, `story_points`, `epic_id`, `priority` (MoSCoW), `dependencies`, `tasks`, `history: []`.
+4. File already present for the **same** `sprint_id`: keep the existing statuses (never move a story back) and only add the new US.
+5. File present for **another** sprint: first move its stories under `archived_sprints.<old sprint_id>` (`status`, `points`), then write the new sprint.
+
+```yaml
+# .bmad/sprint-status.yaml
+version: "1.0"
+metadata:
+  sprint_id: "sprint-2"
+  name: "User management"
+  start_date: "2026-11-02"
+  end_date: "2026-11-13"
+  goal: "Users can sign up, log in and edit their profile"
+  capacity_points: 21
+stories:
+  US-010:
+    title: "User sign-up"
+    status: "ready-for-dev"
+    story_points: 5
+    epic_id: "EPIC-002"
+    priority: "must"
+    dependencies: ["US-001"]
+    tasks: { total: 0, completed: 0, list: [] }
+    history: []
+archived_sprints:
+  sprint-1:
+    stories:
+      US-001: { status: "done", points: 5 }
+```
+
+### Step 7: Final Checklist
 
 ```
 ══════════════════════════════════════════════════════════════
@@ -291,6 +327,9 @@ Points |
 ║                      NEXT STEP                           ║
 ╠══════════════════════════════════════════════════════════╣
 ║                                                          ║
+║  → /gate:validate-sprint {N}                             ║
+║    Check the sprint is ready (gate)                      ║
+║                                                          ║
 ║  → /sprint:next-story                                    ║
 ║    Pick the first story from the sprint backlog          ║
 ║                                                          ║

package/Dev/i18n/es/Common/rules/12-context-management.md

@@ -492,12 +492,14 @@ Durante la compactacion, siempre preservar:
 
 | Variable | Descripción |
 |----------|-------------|
-| `CLAUDE_CODE_SUBAGENT_MODEL` | Modelo **por defecto** para sub-agentes sin tipo (ej: `sonnet`). **El frontmatter `model:` de un agente tiene precedencia**: los 11 reviewers con `model: haiku` permanecen en Haiku aunque esta variable sea `sonnet`. |
+| `CLAUDE_CODE_SUBAGENT_MODEL` | Modelo **por defecto** para sub-agentes sin tipo (ej: `sonnet`). **Desde Claude Code 2.1.251**, el frontmatter `model:` de un agente tiene precedencia: los 11 reviewers con `model: haiku` permanecen en Haiku. **Antes de 2.1.251 era al revés** — esta variable sobrescribía el frontmatter. |
 | `CLAUDE_CODE_FORK_SUBAGENT` | `1` para aislar el contexto de los sub-agentes (forked subagents, v2.1.117+) |
 | `ENABLE_PROMPT_CACHING_1H` / `FORCE_PROMPT_CACHING_5M` | Ampliar/forzar el prompt caching (−40 % en sesiones repetitivas) |
 | `CLAUDE_CODE_DISABLE_AUTO_MEMORY` | Poner a `1` para desactivar la memoria automática |
 
-> **Precedencia de modelos (auditoría 2026-06-08):** `model:` en el frontmatter del agente > `CLAUDE_CODE_SUBAGENT_MODEL` > modelo de sesión. La variable `SUBAGENT_MODEL=sonnet` NO sobreescribe los agentes explícitamente con `model: haiku` — solo aplica a sub-agentes sin `model:` definido.
+> **Precedencia de modelos — válida desde Claude Code 2.1.251:** `model:` pasado en la invocación > `model:` del frontmatter del agente (donde `inherit` toma el modelo de la conversación) > `CLAUDE_CODE_SUBAGENT_MODEL` > modelo de sesión.
+>
+> **⚠️ Antes de 2.1.251 el orden era el inverso:** `CLAUDE_CODE_SUBAGENT_MODEL` iba primero y sobrescribía tanto el parámetro de invocación como el frontmatter, incluido `model: inherit`. En esas versiones, `SUBAGENT_MODEL=sonnet` sí ejecutaba en Sonnet los reviewers `model: haiku` — lo contrario de la intención. Comprueba `claude --version` antes de fiarte de esta tabla. Fuente: [documentación de sub-agents](https://code.claude.com/docs/en/sub-agents).
 
 ---
 

package/Dev/i18n/es/Common/templates/settings.json.template

@@ -18,19 +18,10 @@
       "Write",
       "Glob",
       "Grep",
+      "__STACK_PERMISSIONS__",
       "Bash(git:*)",
-      "Bash(npm run:*)",
-      "Bash(npm test:*)",
-      "Bash(pnpm:*)",
-      "Bash(yarn:*)",
-      "Bash(composer:*)",
-      "Bash(php bin/console:*)",
-      "Bash(php artisan:*)",
-      "Bash(dotnet:*)",
       "Bash(make:*)",
-      "Bash(docker compose:*)",
-      "Bash(flutter:*)",
-      "Bash(ng:*)"
+      "Bash(docker compose:*)"
     ],
     "deny": [
       "Bash(rm -rf:*)",

package/Dev/i18n/es/Workflow/commands/auto-sprint.md

@@ -40,7 +40,7 @@ $ARGUMENTS
 - `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` configurado
 - CLI `gh` autenticada (creación de PR / checks / merge)
 - Docker disponible (todos los tests se ejecutan vía Docker — ver CLAUDE.md del proyecto)
-- Proyecto BMAD v6 con `.bmad/sprint-status.yaml` presente
+- Proyecto BMAD v6: `/workflow:start` (fase 1) crea `.bmad/sprint-status.yaml` si no existe
 
 > Si falta algún requisito previo, abortar de inmediato con un mensaje claro y accionable. No omitir ninguna fase en silencio.
 

package/Dev/i18n/es/Workflow/commands/start.md

@@ -224,7 +224,43 @@ Puntos |
 {Notas del sprint planning, decisiones tomadas...}
 ```
 
-### Paso 6: Lista de Verificación Final
+### Paso 6: Sincronizar `.bmad/sprint-status.yaml`
+
+`/gate:validate-sprint` y los comandos `/sprint:*` leen el sprint en `.bmad/sprint-status.yaml`. Escríbelo ahora a partir del `sprint-goal.md` anterior y de las fichas US seleccionadas (`project-management/backlog/user-stories/US-*.md`):
+
+1. Crear `.bmad/` si no existe.
+2. `metadata`: `sprint_id` (`sprint-{N}`), `name`, `start_date`, `end_date`, `goal`, `capacity_points`, tomados de `sprint-goal.md`.
+3. `stories`: una entrada por US seleccionada: `title`, `status: ready-for-dev`, `story_points`, `epic_id`, `priority` (MoSCoW), `dependencies`, `tasks`, `history: []`.
+4. Archivo ya presente para el **mismo** `sprint_id`: conservar los estados existentes (nunca hacer retroceder una story) y añadir solo las nuevas US.
+5. Archivo presente para **otro** sprint: mover primero sus stories a `archived_sprints.<sprint_id anterior>` (`status`, `points`) y luego escribir el nuevo sprint.
+
+```yaml
+# .bmad/sprint-status.yaml
+version: "1.0"
+metadata:
+  sprint_id: "sprint-2"
+  name: "Gestión de usuarios"
+  start_date: "2026-11-02"
+  end_date: "2026-11-13"
+  goal: "Los usuarios pueden registrarse, iniciar sesión y editar su perfil"
+  capacity_points: 21
+stories:
+  US-010:
+    title: "Registro de usuario"
+    status: "ready-for-dev"
+    story_points: 5
+    epic_id: "EPIC-002"
+    priority: "must"
+    dependencies: ["US-001"]
+    tasks: { total: 0, completed: 0, list: [] }
+    history: []
+archived_sprints:
+  sprint-1:
+    stories:
+      US-001: { status: "done", points: 5 }
+```
+
+### Paso 7: Lista de Verificación Final
 
 ```
 ══════════════════════════════════════════════════════════════
@@ -291,6 +327,9 @@ Puntos |
 ║                    SIGUIENTE PASO                        ║
 ╠══════════════════════════════════════════════════════════╣
 ║                                                          ║
+║  → /gate:validate-sprint {N}                             ║
+║    Verificar que el sprint está listo (gate)             ║
+║                                                          ║
 ║  → /sprint:next-story                                    ║
 ║    Tomar la primera story del sprint backlog             ║
 ║                                                          ║

package/Dev/i18n/fr/Common/rules/12-context-management.md

@@ -492,12 +492,14 @@ Lors de la compaction, toujours preserver:
 
 | Variable | Description |
 |----------|-------------|
-| `CLAUDE_CODE_SUBAGENT_MODEL` | Modèle **par défaut** des sous-agents non typés (ex : `sonnet`). **Le frontmatter `model:` d'un agent prévaut** : les 11 reviewers `model: haiku` restent sur Haiku même avec cette variable à `sonnet`. |
+| `CLAUDE_CODE_SUBAGENT_MODEL` | Modèle **par défaut** des sous-agents non typés (ex : `sonnet`). **Depuis Claude Code 2.1.251**, le frontmatter `model:` d'un agent prévaut : les 11 reviewers `model: haiku` restent sur Haiku. **Avant 2.1.251, c'était l'inverse** — cette variable écrasait le frontmatter. |
 | `CLAUDE_CODE_FORK_SUBAGENT` | `1` pour isoler le contexte des sous-agents (forked subagents, v2.1.117+) |
 | `ENABLE_PROMPT_CACHING_1H` / `FORCE_PROMPT_CACHING_5M` | Étendre/forcer le prompt caching (−40 % sur les sessions répétitives) |
 | `CLAUDE_CODE_DISABLE_AUTO_MEMORY` | Mettre à `1` pour désactiver la mémoire automatique |
 
-> **Précédence modèle (clarification audit 2026-06-08) :** `model:` dans le frontmatter d'un agent > `CLAUDE_CODE_SUBAGENT_MODEL` > modèle de la session. La variable `SUBAGENT_MODEL=sonnet` n'écrase donc PAS les agents explicitement `model: haiku` — elle ne s'applique qu'aux sous-agents sans `model:` défini.
+> **Précédence modèle — valable depuis Claude Code 2.1.251 :** `model:` passé à l'invocation > `model:` du frontmatter de l'agent (où `inherit` reprend le modèle de la conversation) > `CLAUDE_CODE_SUBAGENT_MODEL` > modèle de la session.
+>
+> **⚠️ Avant 2.1.251, l'ordre était inversé :** `CLAUDE_CODE_SUBAGENT_MODEL` arrivait en tête et écrasait à la fois le paramètre d'invocation et le frontmatter, y compris `model: inherit`. Sur ces versions, `SUBAGENT_MODEL=sonnet` faisait bel et bien tourner sur Sonnet les reviewers `model: haiku` — l'inverse de l'intention. Vérifier `claude --version` avant de se fier à cette table. Source : [doc sub-agents](https://code.claude.com/docs/en/sub-agents).
 
 ---
 

package/Dev/i18n/fr/Common/templates/settings.json.template

@@ -18,19 +18,10 @@
       "Write",
       "Glob",
       "Grep",
+      "__STACK_PERMISSIONS__",
       "Bash(git:*)",
-      "Bash(npm run:*)",
-      "Bash(npm test:*)",
-      "Bash(pnpm:*)",
-      "Bash(yarn:*)",
-      "Bash(composer:*)",
-      "Bash(php bin/console:*)",
-      "Bash(php artisan:*)",
-      "Bash(dotnet:*)",
       "Bash(make:*)",
-      "Bash(docker compose:*)",
-      "Bash(flutter:*)",
-      "Bash(ng:*)"
+      "Bash(docker compose:*)"
     ],
     "deny": [
       "Bash(rm -rf:*)",

package/Dev/i18n/fr/Workflow/commands/auto-sprint.md

@@ -40,7 +40,7 @@ $ARGUMENTS
 - `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` défini
 - CLI `gh` authentifié (création de PR / checks / merge)
 - Docker disponible (tous les tests s'exécutent via Docker — voir le CLAUDE.md du projet)
-- Projet BMAD v6 avec `.bmad/sprint-status.yaml` présent
+- Projet BMAD v6 : `.bmad/sprint-status.yaml` est créé par `/workflow:start` (phase 1) s'il n'existe pas
 
 > Si un prérequis est manquant, abandonner immédiatement avec un message clair et actionnable. Ne jamais ignorer silencieusement une phase.
 

package/Dev/i18n/fr/Workflow/commands/start.md

@@ -224,7 +224,43 @@ Points |
 {Notes de sprint planning, décisions prises...}
 ```
 
-### Étape 6 : Checklist Finale
+### Étape 6 : Synchroniser `.bmad/sprint-status.yaml`
+
+`/gate:validate-sprint` et les commandes `/sprint:*` lisent le sprint dans `.bmad/sprint-status.yaml`. L'écrire maintenant à partir du `sprint-goal.md` ci-dessus et des fiches US sélectionnées (`project-management/backlog/user-stories/US-*.md`) :
+
+1. Créer `.bmad/` s'il n'existe pas.
+2. `metadata` : `sprint_id` (`sprint-{N}`), `name`, `start_date`, `end_date`, `goal`, `capacity_points`, repris de `sprint-goal.md`.
+3. `stories` : une entrée par US sélectionnée : `title`, `status: ready-for-dev`, `story_points`, `epic_id`, `priority` (MoSCoW), `dependencies`, `tasks`, `history: []`.
+4. Fichier déjà présent pour le **même** `sprint_id` : conserver les statuts existants (ne jamais faire reculer une story) et n'ajouter que les nouvelles US.
+5. Fichier présent pour un **autre** sprint : déplacer d'abord ses stories sous `archived_sprints.<ancien sprint_id>` (`status`, `points`), puis écrire le nouveau sprint.
+
+```yaml
+# .bmad/sprint-status.yaml
+version: "1.0"
+metadata:
+  sprint_id: "sprint-2"
+  name: "Gestion des utilisateurs"
+  start_date: "2026-11-02"
+  end_date: "2026-11-13"
+  goal: "Les utilisateurs peuvent s'inscrire, se connecter et modifier leur profil"
+  capacity_points: 21
+stories:
+  US-010:
+    title: "Inscription utilisateur"
+    status: "ready-for-dev"
+    story_points: 5
+    epic_id: "EPIC-002"
+    priority: "must"
+    dependencies: ["US-001"]
+    tasks: { total: 0, completed: 0, list: [] }
+    history: []
+archived_sprints:
+  sprint-1:
+    stories:
+      US-001: { status: "done", points: 5 }
+```
+
+### Étape 7 : Checklist Finale
 
 ```
 ══════════════════════════════════════════════════════════════
@@ -291,6 +327,9 @@ Points |
 ║                   PROCHAINE ÉTAPE                        ║
 ╠══════════════════════════════════════════════════════════╣
 ║                                                          ║
+║  → /gate:validate-sprint {N}                             ║
+║    Vérifier que le sprint est prêt (gate)                ║
+║                                                          ║
 ║  → /sprint:next-story                                    ║
 ║    Prendre la première story du sprint backlog           ║
 ║                                                          ║

package/Dev/i18n/pt/Common/rules/12-context-management.md

@@ -492,12 +492,14 @@ Durante a compactacao, sempre preservar:
 
 | Variável | Descrição |
 |----------|-----------|
-| `CLAUDE_CODE_SUBAGENT_MODEL` | Modelo **padrão** para sub-agentes sem tipo (ex: `sonnet`). **O frontmatter `model:` de um agente tem prioridade**: 11 revisores com `model: haiku` permanecem no Haiku mesmo com esta variável em `sonnet`. |
+| `CLAUDE_CODE_SUBAGENT_MODEL` | Modelo **padrão** para sub-agentes sem tipo (ex: `sonnet`). **Desde o Claude Code 2.1.251**, o frontmatter `model:` de um agente tem prioridade: os 11 revisores com `model: haiku` permanecem no Haiku. **Antes de 2.1.251 era o contrário** — esta variável sobrescrevia o frontmatter. |
 | `CLAUDE_CODE_FORK_SUBAGENT` | `1` para isolar o contexto dos sub-agentes (forked subagents, v2.1.117+) |
 | `ENABLE_PROMPT_CACHING_1H` / `FORCE_PROMPT_CACHING_5M` | Ampliar/forçar o prompt caching (−40 % em sessões repetitivas) |
 | `CLAUDE_CODE_DISABLE_AUTO_MEMORY` | Definir como `1` para desativar a memória automática |
 
-> **Precedência de modelos (auditoria 2026-06-08):** `model:` no frontmatter do agente > `CLAUDE_CODE_SUBAGENT_MODEL` > modelo de sessão. A variável `SUBAGENT_MODEL=sonnet` NÃO sobrescreve agentes explicitamente com `model: haiku` — aplica-se apenas a sub-agentes sem `model:` definido.
+> **Precedência de modelos — válida desde o Claude Code 2.1.251:** `model:` passado na invocação > `model:` do frontmatter do agente (onde `inherit` assume o modelo da conversa) > `CLAUDE_CODE_SUBAGENT_MODEL` > modelo de sessão.
+>
+> **⚠️ Antes de 2.1.251 a ordem era inversa:** `CLAUDE_CODE_SUBAGENT_MODEL` vinha primeiro e sobrescrevia tanto o parâmetro de invocação quanto o frontmatter, incluindo `model: inherit`. Nessas versões, `SUBAGENT_MODEL=sonnet` de facto executava em Sonnet os revisores `model: haiku` — o oposto da intenção. Verifique `claude --version` antes de confiar nesta tabela. Fonte: [documentação de sub-agents](https://code.claude.com/docs/en/sub-agents).
 
 ---
 

package/Dev/i18n/pt/Common/templates/settings.json.template

@@ -18,19 +18,10 @@
       "Write",
       "Glob",
       "Grep",
+      "__STACK_PERMISSIONS__",
       "Bash(git:*)",
-      "Bash(npm run:*)",
-      "Bash(npm test:*)",
-      "Bash(pnpm:*)",
-      "Bash(yarn:*)",
-      "Bash(composer:*)",
-      "Bash(php bin/console:*)",
-      "Bash(php artisan:*)",
-      "Bash(dotnet:*)",
       "Bash(make:*)",
-      "Bash(docker compose:*)",
-      "Bash(flutter:*)",
-      "Bash(ng:*)"
+      "Bash(docker compose:*)"
     ],
     "deny": [
       "Bash(rm -rf:*)",

package/Dev/i18n/pt/Workflow/commands/auto-sprint.md

@@ -40,7 +40,7 @@ $ARGUMENTS
 - `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1` definido
 - CLI `gh` autenticado (criação de PR / verificações / merge)
 - Docker disponível (todos os testes são executados via Docker — veja o CLAUDE.md do projeto)
-- Projeto BMAD v6 com `.bmad/sprint-status.yaml` presente
+- Projeto BMAD v6: `.bmad/sprint-status.yaml` é criado por `/workflow:start` (fase 1) se não existir
 
 > Se um pré-requisito estiver ausente, aborte imediatamente com uma mensagem clara e acionável. Não pule uma fase silenciosamente.
 

package/Dev/i18n/pt/Workflow/commands/start.md

@@ -224,7 +224,43 @@ Pontos |
 {Notas do sprint planning, decisões tomadas...}
 ```
 
-### Etapa 6: Checklist Final
+### Etapa 6: Sincronizar `.bmad/sprint-status.yaml`
+
+`/gate:validate-sprint` e os comandos `/sprint:*` leem o sprint em `.bmad/sprint-status.yaml`. Escreva-o agora a partir do `sprint-goal.md` acima e das fichas US selecionadas (`project-management/backlog/user-stories/US-*.md`):
+
+1. Criar `.bmad/` se não existir.
+2. `metadata`: `sprint_id` (`sprint-{N}`), `name`, `start_date`, `end_date`, `goal`, `capacity_points`, retirados de `sprint-goal.md`.
+3. `stories`: uma entrada por US selecionada: `title`, `status: ready-for-dev`, `story_points`, `epic_id`, `priority` (MoSCoW), `dependencies`, `tasks`, `history: []`.
+4. Arquivo já presente para o **mesmo** `sprint_id`: manter os status existentes (nunca fazer uma story retroceder) e adicionar apenas as novas US.
+5. Arquivo presente para **outro** sprint: mover primeiro as suas stories para `archived_sprints.<sprint_id anterior>` (`status`, `points`) e depois escrever o novo sprint.
+
+```yaml
+# .bmad/sprint-status.yaml
+version: "1.0"
+metadata:
+  sprint_id: "sprint-2"
+  name: "Gestão de usuários"
+  start_date: "2026-11-02"
+  end_date: "2026-11-13"
+  goal: "Os usuários podem se cadastrar, entrar e editar o seu perfil"
+  capacity_points: 21
+stories:
+  US-010:
+    title: "Cadastro de usuário"
+    status: "ready-for-dev"
+    story_points: 5
+    epic_id: "EPIC-002"
+    priority: "must"
+    dependencies: ["US-001"]
+    tasks: { total: 0, completed: 0, list: [] }
+    history: []
+archived_sprints:
+  sprint-1:
+    stories:
+      US-001: { status: "done", points: 5 }
+```
+
+### Etapa 7: Checklist Final
 
 ```
 ══════════════════════════════════════════════════════════════
@@ -291,6 +327,9 @@ Pontos |
 ║                    PRÓXIMO PASSO                         ║
 ╠══════════════════════════════════════════════════════════╣
 ║                                                          ║
+║  → /gate:validate-sprint {N}                             ║
+║    Verificar se o sprint está pronto (gate)              ║
+║                                                          ║
 ║  → /sprint:next-story                                    ║
 ║    Pegar a primeira story do sprint backlog              ║
 ║                                                          ║