sincron-auto 1.0.0

package/docs/template-permissoes.json

@@ -0,0 +1,83 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "$comment": "Templates de permissões para diferentes níveis de automação",
+
+  "automacao_total": {
+    "permissions": {
+      "allow": [
+        "Read",
+        "Write",
+        "Edit",
+        "Glob",
+        "Grep",
+        "Bash(git *)",
+        "Bash(npm *)",
+        "Bash(npx *)",
+        "Bash(node *)",
+        "Bash(pip *)",
+        "Bash(python *)",
+        "Bash(pytest *)",
+        "Bash(cargo *)",
+        "Bash(go *)",
+        "Bash(dotnet *)",
+        "Bash(mkdir *)",
+        "Bash(cp *)",
+        "Bash(mv *)",
+        "Bash(rm *)",
+        "Bash(touch *)",
+        "Bash(cat *)",
+        "Bash(ls *)",
+        "Bash(pwd)",
+        "Bash(echo *)",
+        "Bash(curl *)",
+        "Bash(wget *)",
+        "Bash(tar *)",
+        "Bash(unzip *)",
+        "Bash(powershell *)",
+        "Bash(cmd *)",
+        "Bash(code *)",
+        "Bash(claude *)",
+        "WebFetch",
+        "WebSearch",
+        "Task",
+        "Skill(*)",
+        "mcp__playwright__*",
+        "mcp__supabase__*",
+        "mcp__ide__*",
+        "mcp__context7__*",
+        "mcp__gscServer__*",
+        "NotebookEdit",
+        "AskUserQuestion"
+      ],
+      "deny": []
+    }
+  },
+
+  "automacao_parcial": {
+    "$comment": "Apenas ferramentas core do Claude",
+    "permissions": {
+      "allow": [
+        "Read",
+        "Write",
+        "Edit",
+        "Glob",
+        "Grep",
+        "WebFetch",
+        "WebSearch",
+        "Task",
+        "Skill(*)",
+        "NotebookEdit",
+        "AskUserQuestion"
+      ],
+      "deny": []
+    }
+  },
+
+  "automacao_nenhuma": {
+    "$comment": "Mantém permissões existentes no settings.local.json sem incluir novas",
+    "permissions": {
+      "allow": [],
+      "deny": []
+    }
+  }
+}

package/package.json

@@ -0,0 +1,34 @@
+{
+  "name": "sincron-auto",
+  "version": "1.0.0",
+  "description": "Plugin de automação multi-agente para Claude Code",
+  "bin": {
+    "sincron-auto": "bin/install.js"
+  },
+  "files": [
+    "bin",
+    "commands",
+    "agents",
+    "skills",
+    "templates",
+    "docs"
+  ],
+  "keywords": [
+    "automation",
+    "multi-agent",
+    "claude-code",
+    "plugin",
+    "development",
+    "workflow"
+  ],
+  "author": "Sincron Framework",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/MLTCorp/sincron-auto"
+  },
+  "homepage": "https://github.com/MLTCorp/sincron-auto#readme",
+  "engines": {
+    "node": ">=14.0.0"
+  }
+}

package/skills/autorizacao/SKILL.md

@@ -0,0 +1,78 @@
+---
+name: autorizacao
+description: Configure project permissions in settings.local.json based on automation level from .sincron-auto/request.json.
+---
+
+# Autorizacao Skill
+
+Configure Claude Code permissions based on automation level.
+
+## When to Use
+
+ONLY during first-run setup by the Estrutura agent.
+
+## Process
+
+### Step 1: Read Automation Level
+
+```
+Read .sincron-auto/request.json → automation_level
+```
+
+Map to level:
+- "Total (Recomendado)" or "Total" → Total
+- "Parcial" → Parcial
+- "Nenhuma" → Nenhuma
+
+### Step 2: Read Current Settings
+
+```
+Read .claude/settings.local.json
+```
+
+If file doesn't exist, create initial structure:
+```json
+{ "permissions": { "allow": [] } }
+```
+
+### Step 3: Select and Read Template
+
+**If Total:**
+```
+Read templates/settings-total.json
+```
+
+**If Parcial:**
+```
+Read templates/settings-parcial.json
+```
+
+**If Nenhuma:**
+→ Skip to Step 5 (keep existing)
+
+### Step 4: Merge Permissions
+
+Merge template permissions into current settings:
+```
+newPermissions = [...currentPermissions, ...templatePermissions]
+Remove duplicates
+```
+
+### Step 5: Write Updated Settings
+
+```
+Write .claude/settings.local.json with merged permissions
+```
+
+### Step 6: Confirm
+
+```
+✅ Permissões configuradas!
+Nível: [level]
+```
+
+## Constraints
+
+- NEVER use AskUserQuestion (read level from file)
+- NEVER overwrite existing permissions (merge only)
+- ONLY modify permissions.allow array

package/skills/avaliacao/SKILL.md

@@ -0,0 +1,55 @@
+---
+name: avaliacao
+description: This skill should be used to perform comprehensive functional and visual validation using Playwright MCP. Validates all acceptance criteria of a task and maintains evaluation report.
+---
+
+# Avaliacao Skill
+
+This skill performs comprehensive testing and validation against acceptance criteria.
+
+## Process
+
+### Step 1: Receive Input
+
+You receive:
+### Step 2: Read Files
+
+Read all files_changed to understand implementation.
+
+### Step 3: Functional Testing
+
+Test each functional criterion:
+- API endpoints
+- Logic correctness
+- Error handling
+
+### Step 4: Visual Testing (if applicable)
+
+Use Playwright MCP for UI criteria:
+### Step 5: Collect Issues
+
+For each failed criterion, document:
+### Step 6: Update Report
+
+Append to relatorio-avaliacao.json:
+### Step 7: Check Attempt Limit (CRITICAL - PREVENTS DEADLOCK)
+
+**BEFORE deciding result**, check limit:
+
+### Step 8: Make Decision
+
+**If limit NOT reached**:
+
+- ALL criteria passed → - ANY critical/high issues → - Multiple medium issues → 
+**If limit reached**:
+- \ (regardless of criteria)
+
+### Step 9: Return Result
+
+Return to Avaliador agent.
+
+## Important Notes
+
+- ALWAYS check limit BEFORE deciding
+- NEVER allow infinite loops (limit prevents deadlock)
+- ONLY modify relatorio-avaliacao.json

package/skills/gerenciar/SKILL.md

@@ -0,0 +1,60 @@
+---
+name: gerenciar
+description: This skill should be used for FIFO task queue management. Lists pending tasks, identifies next task by execution order, updates status, and returns task for Construtor.
+---
+
+# Gerenciar Skill
+
+This skill manages the FIFO task queue and selects the next task for execution.
+
+## When to Use
+
+Use this skill to get the next pending task in FIFO order.
+
+## Process
+
+### Step 1: Read State
+
+Read `sincron-auto-state.json` to get all tasks.
+
+### Step 2: Filter Pending Tasks
+
+Filter tasks where:
+- `status === "pendente"`
+- `abandoned !== true`
+
+### Step 3: Sort by Execution Order
+
+Sort filtered tasks by `execution_order` (ascending).
+
+### Step 4: Select First Task
+
+Select the FIRST task from sorted list (FIFO).
+
+If no tasks match criteria, return:
+```json
+{
+  "status": "no_pending_tasks"
+}
+```
+
+### Step 5: Update Task Status
+
+Update selected task in state:
+```json
+{
+  "status": "em execução"
+}
+```
+
+Save state file.
+
+### Step 6: Return Task
+
+Return the task object to Gestor Projeto for delegation to Construtor.
+
+## Important Notes
+
+- STRICT FIFO: Always select task with lowest execution_order
+- NEVER skip tasks (unless abandoned)
+- ONE task at a time

package/skills/instalar-mcp/SKILL.md

@@ -0,0 +1,59 @@
+---
+name: instalar-mcp
+description: Install required MCP servers (Context7 and Playwright). Use during first-run setup.
+---
+
+# Instalar MCP Skill
+
+Install Context7 and Playwright MCPs for Sincron-Auto.
+
+## Process
+
+### Step 1: Check Existing MCPs
+
+```bash
+claude mcp list
+```
+
+Check if context7 and playwright are already installed.
+
+### Step 2: Install Context7 (if needed)
+
+```bash
+claude mcp add --transport http context7 https://mcp.context7.com/mcp --header "Header-Name: ctx7sk-f1688803-3139-4a02-93df-55bf946be647"
+```
+
+### Step 3: Install Playwright (if needed)
+
+```bash
+claude mcp add playwright npx @playwright/mcp@latest
+```
+
+### Step 4: Verify
+
+```bash
+claude mcp list
+```
+
+Confirm both MCPs appear in the list.
+
+### Step 5: Report
+
+**Success:**
+```
+✅ MCPs instalados:
+- Context7 ✓
+- Playwright ✓
+```
+
+**Failure:**
+```
+❌ Falha ao instalar MCPs:
+[show manual commands]
+```
+
+## Constraints
+
+- ONLY install Context7 and Playwright
+- DO NOT reinstall if already present
+- Provide manual commands if automatic install fails

package/skills/projetar/SKILL.md

@@ -0,0 +1,53 @@
+---
+name: projetar
+description: This skill should be used when creating or reviewing project planning. It analyzes requirements to create user stories and breaks them down into granular tasks with specific acceptance criteria.
+---
+
+# Projetar Skill
+
+This skill transforms user requirements into structured planning: user stories → tasks → acceptance criteria.
+
+## When to Use
+
+- Creating new project (first-time planning)
+- Adding new features to existing project
+- Reviewing/adjusting project scope
+
+## Process
+
+### Step 1: Analyze Requirements
+
+Extract key functionality from user request.
+
+### Step 2: Create User Stories
+
+Format: "Como [user], eu quero [action] para [benefit]"
+
+### Step 3: Break Into Tasks
+
+One user story can generate multiple tasks. Each task should be granular and focused.
+
+### Step 4: Define Acceptance Criteria
+
+Each task gets multiple specific, testable criteria.
+
+### Step 4.5: Classify Criteria (SMART CRITERIA - PREVENTS DEADLOCK)
+
+For each criterion, optionally classify as:
+
+- **isolated**: Testable with ONLY this task complete
+- **dependent**: Requires OTHER tasks complete
+- **integration**: Tests interaction between components
+
+Guidelines:
+- Early tasks → prefer isolated criteria
+- Late tasks → can have dependent/integration criteria
+- Create separate integration task at end
+
+### Step 5: Order Tasks
+
+Set execution_order field based on dependencies.
+
+### Step 6: Return Tasks Array
+
+Return structured tasks for Gestor Projeto to save in state.

package/skills/relatorio-final/SKILL.md

@@ -0,0 +1,269 @@
+---
+name: relatorio-final
+description: This skill should be used to compile a comprehensive user-friendly markdown report from technical evaluation data. Only use when Gestor Projeto notifies that all tasks are complete.
+---
+
+# Relatorio Final Skill
+
+This skill generates a user-friendly markdown report from the technical state and evaluation data after all tasks are processed.
+
+## When to Use
+
+Use this skill ONLY when:
+- All tasks have status `completo` OR `not_approved`
+- No tasks remain with status `pendente` or `em execução`
+- Gestor Projeto has notified completion
+
+DO NOT use this skill:
+- During task execution
+- When tasks are still pending
+- For intermediate progress reports
+
+## Process
+
+### Step 1: Read State Files
+
+Read the following files:
+
+1. `sincron-auto-state.json` - Project and task state
+2. `relatorio-avaliacao.json` - Evaluation history
+
+Verify both files exist and have valid JSON.
+
+### Step 2: Extract Key Metrics
+
+From `sincron-auto-state.json`, calculate:
+
+```javascript
+const totalTasks = state.tasks.length
+const approvedTasks = state.tasks.filter(t => t.status === 'completo').length
+const notApprovedTasks = state.tasks.filter(t => t.status === 'not_approved').length
+const abandonedTasks = state.tasks.filter(t => t.abandoned === true).length
+```
+
+From `relatorio-avaliacao.json`, count:
+
+```javascript
+const totalEvaluations = evaluations.length
+const rejections = evaluations.filter(e => e.result === 'rejected').length
+const approvals = evaluations.filter(e => e.result === 'approved').length
+```
+
+### Step 3: Organize Content
+
+Structure the report with the following sections:
+
+1. **Header** - Project name, date, summary metrics
+2. **Executive Summary** - High-level overview
+3. **Tasks Aprovadas** - Successfully completed tasks
+4. **Tasks Não Aprovadas** - Tasks that hit attempt limit (if any)
+5. **Tasks Abandonadas** - User-abandoned tasks (if any)
+6. **Histórico de Avaliações** - Evaluation timeline
+7. **Recomendações** - Next steps based on results
+
+### Step 4: Generate Markdown Content
+
+Create markdown with this structure:
+
+```markdown
+# Relatório Final - [Project Name]
+
+**Data**: [YYYY-MM-DD HH:MM]
+**Nível de Automação**: [total|parcial|nenhuma]
+
+---
+
+## Resumo Executivo
+
+Este projeto consistiu em [total] tasks relacionadas a: [brief project description from user stories].
+
+### Métricas Gerais
+
+- ✅ **Tasks Aprovadas**: X/Y (Z%)
+- ⚠️ **Tasks Não Aprovadas**: A/Y (B%)
+- 🚫 **Tasks Abandonadas**: C/Y (D%)
+- 🔄 **Total de Avaliações**: E
+- 📊 **Taxa de Sucesso na Primeira Tentativa**: F%
+
+---
+
+## Tasks Aprovadas ✅
+
+[For each approved task, create subsection]
+
+### [Task ID]: [Task Name]
+
+**User Story**: [Associated user story]
+
+**Critérios Atendidos**:
+- [Criterion 1]
+- [Criterion 2]
+- [...]
+
+**Tentativas até Aprovação**: X
+
+**Arquivos Modificados**:
+- [file1.js]
+- [file2.css]
+
+---
+
+## Tasks Não Aprovadas ⚠️
+
+[If no tasks with status not_approved, show: "Nenhuma task ficou não aprovada."]
+
+[For each not_approved task]
+
+### [Task ID]: [Task Name]
+
+**User Story**: [Associated user story]
+
+**Tentativas**: [evaluation_attempts]/[max_evaluation_attempts]
+
+**Motivo**: [not_approved_reason]
+
+**Critérios que Falharam**:
+- [Failed criterion 1]
+- [Failed criterion 2]
+
+**Último Issues Encontrados**:
+
+[From last evaluation in relatorio-avaliacao.json for this task]
+
+1. **[Issue Type]** ([Severity])
+   - **Descrição**: [Issue description]
+   - **Recomendação**: [Issue recommendation]
+   - **Critério Relacionado**: [related_criteria]
+
+**O Que Fazer Agora**:
+
+Based on the issues, suggest one or more:
+
+- ✏️ **Investigar Manualmente**: [Specific files/areas to check]
+- 🔄 **Reexecutar**: Use `/sincron-auto continuar` para tentar novamente
+- 📋 **Replanejar**: Considere dividir em tasks menores
+- ✅ **Aceitar Como Está**: Se não for crítico, pode seguir em frente
+
+---
+
+## Tasks Abandonadas 🚫
+
+[If no abandoned tasks, show: "Nenhuma task foi abandonada."]
+
+[For each abandoned task]
+
+### [Task ID]: [Task Name]
+
+**Motivo do Abandono**: [User requested abandonment]
+
+**User Story Original**: [Associated user story]
+
+---
+
+## Histórico de Avaliações 📋
+
+Timeline completo de todas as avaliações realizadas:
+
+[For each evaluation in relatorio-avaliacao.json, chronologically]
+
+**[Timestamp]** - Task [task_id]: [result]
+- Tentativa: X/Y
+- Issues encontrados: Z
+- [Brief summary]
+
+---
+
+## Recomendações 💡
+
+Based on the overall results, provide actionable recommendations:
+
+### Próximos Passos Sugeridos
+
+1. **Se Todas Tasks Aprovadas**:
+   - ✅ Projeto pronto para testes manuais
+   - ✅ Considere executar testes de integração end-to-end
+   - ✅ Deploy para ambiente de staging/produção
+
+2. **Se Há Tasks Não Aprovadas**:
+   - ⚠️ Revisar tasks não aprovadas (veja seção acima)
+   - ⚠️ Investigar issues específicos listados
+   - ⚠️ Considere replanejar tasks problemáticas
+   - ⚠️ Ou aceite como não-crítico e prossiga
+
+3. **Se Há Tasks Abandonadas**:
+   - 🚫 Avaliar se funcionalidade é realmente desnecessária
+   - 🚫 Ou retome com `/sincron-auto adicionar [feature]` se mudou de ideia
+
+### Melhorias Futuras
+
+[Based on patterns in failed tasks, suggest improvements]
+
+Exemplo:
+- Se múltiplas tasks falharam em testes visuais: "Considere revisar design system"
+- Se múltiplas tasks falharam em integração: "Considere task dedicada a integração end-to-end"
+
+---
+
+## Arquivos Gerados
+
+Este relatório foi gerado automaticamente pelo Sincron-Auto.
+
+**Arquivos de referência**:
+- Estado do projeto: `sincron-auto-state.json`
+- Histórico de avaliações: `relatorio-avaliacao.json`
+- Este relatório: `reports/relatorio-[project]-[date].md`
+
+---
+
+**Gerado por**: Sincron-Auto v1.0.0
+**Data**: [YYYY-MM-DD HH:MM:SS]
+```
+
+### Step 5: Save Report
+
+Generate filename with format:
+```
+reports/relatorio-[project-name]-[YYYY-MM-DD-HHMM].md
+```
+
+Sanitize project name (remove spaces, special chars, lowercase).
+
+Write the markdown content to the file.
+
+### Step 6: Return Summary
+
+Return a brief summary to Orquestrador:
+
+```json
+{
+  "report_path": "reports/relatorio-[...].md",
+  "total_tasks": X,
+  "approved_tasks": Y,
+  "not_approved_tasks": Z,
+  "abandoned_tasks": A,
+  "success_rate": "B%"
+}
+```
+
+## Important Notes
+
+- DO NOT modify any source files (read-only skill)
+- DO NOT use overly technical language in report (user-friendly)
+- ALWAYS include actionable recommendations
+- NEVER skip not_approved tasks section (critical for user)
+- Use emojis sparingly for visual organization
+- Keep tone professional but accessible
+- Focus on WHAT happened and WHAT TO DO NEXT
+
+## Example Output Path
+
+```
+reports/relatorio-sistema-blog-2026-01-26-1430.md
+```
+
+## Constraints
+
+- Maximum report length: No strict limit, but aim for clarity over length
+- ALWAYS include all sections even if empty
+- NEVER omit not_approved tasks (most important for user)
+- DO NOT make assumptions about why tasks failed (use actual evaluation data)

package/templates/settings-nenhuma.json

@@ -0,0 +1,5 @@
+{
+  "permissions": {
+    "allow": []
+  }
+}