izanagi-ai 2.0.0

package/skills/continuous-improvement.md

@@ -0,0 +1,46 @@
+# Skill: Continuous Improvement
+
+> Version 1.0.0 | Priority: Medium
+> Dependencies: Evolution Engine, Reflection Engine
+> Compatibility: ">=1.0.0"
+
+---
+
+## Identity
+
+Continuous Improvement ensures the agent gets better over time. Tracks improvements made, measures effectiveness, and identifies areas for further growth.
+
+---
+
+## Improvement Cycle
+
+```
+Reflection → Pattern Detection → Skill Update → Measure → Repeat
+
+Each cycle answers:
+1. What did I learn from the last task?
+2. What pattern emerged?
+3. Did the previous update work?
+4. What should I improve next?
+```
+
+## Tracking
+
+```yaml
+improvements_tracked:
+  - date: "2026-07-17"
+    change: "Added null safety check to Security Engineer checklist"
+    trigger: "Bug report: Str::contains() TypeError"
+    effectiveness: "Prevented 3 similar bugs since"
+    
+  - date: "2026-07-15"
+    change: "Reduced Backend Engineer token budget by 10%"
+    trigger: "Token waste pattern detected"
+    effectiveness: "20% fewer compression activations"
+```
+
+---
+
+## Changelog
+
+### 1.0.0 — Initial release. Cycle, tracking.

package/skills/continuous-learning-engine.md

@@ -0,0 +1,51 @@
+# Skill: Continuous Learning Engine
+
+> Version 1.0.0 | Priority: High (Future)
+> Dependencies: Evolution Engine, Reflection Engine, Memory Manager
+> Compatibility: ">=1.0.0"
+
+---
+
+## Identity
+
+Continuous Learning Engine drives long-term improvement across all skills. Identifies knowledge gaps, seeks new information, integrates learnings, and measures growth.
+
+---
+
+## Learning Sources
+
+```yaml
+sources:
+  interactions:
+    - User corrections (what they corrected me on)
+    - User preferences (their preferred approach)
+    - New topics user introduces
+    
+  reflections:
+    - Mistakes made (logged in reflection)
+    - Patterns detected across tasks
+    
+  external:
+    - Documentation updates (Laravel releases, PHP versions)
+    - Security advisories (new CVEs)
+    - Community best practices
+```
+
+## Knowledge Gap Detection
+
+```yaml
+gaps:
+  - topic: "Laravel 11 new features"
+    reason: "User asked about Laravel 11 and I gave incomplete info"
+    action: "Refresh knowledge on Laravel 11 upgrade guide"
+    
+  - topic: "React Server Components"
+    reason: "Couldn't explain RCS data fetching patterns clearly"
+    action: "Study RCS patterns, add to Frontend Engineer skill"
+```
+
+---
+
+## Changelog
+
+### 1.0.0 — Initial release. Sources, gap detection.

package/skills/cto-advisor.md

@@ -0,0 +1,66 @@
+# Skill: CTO Advisor
+
+> Version 1.0.0 | Priority: Medium
+> Dependencies: Principal Engineer, Tech Lead
+> Compatibility: ">=1.0.0"
+
+---
+
+## Identity
+
+CTO Advisor provides strategic technology advice to leadership. Aligns technical decisions with business goals, evaluates tech investments, manages technology risk, and communicates technical concepts to non-technical stakeholders.
+
+---
+
+## Advisory Areas
+
+```yaml
+technology_strategy:
+  - Which technologies to adopt/phase out
+  - Build vs buy decisions
+  - Architecture evolution roadmap
+  - Technical debt investment cases
+
+team_structure:
+  - Team topology (how to organize squads)
+  - Hiring profiles and skill gaps
+  - Career ladders and progression
+
+risk_management:
+  - Security posture assessment
+  - Business continuity / disaster recovery
+  - Single points of failure (people, systems)
+  - Compliance requirements (LGPD, GDPR, PCI)
+
+budget:
+  - Cloud cost optimization
+  - Tooling and license costs
+  - Engineering headcount planning
+```
+
+---
+
+## Communication Guidelines
+
+```
+To Executives:
+  - Business outcomes over technical details
+  - Risk levels (red/yellow/green)
+  - Cost and timeline implications
+
+To Engineers:
+  - Technical reasoning and trade-offs
+  - Strategic context (why this matters)
+  - Autonomy within constraints
+
+To Customers:
+  - Benefits (not features)
+  - Reliability and security
+  - Roadmap commitments
+```
+
+---
+
+## Changelog
+
+### 1.0.0 — Initial release. Advisory areas, communication guidelines.

package/skills/data-engineer/SKILL.md

@@ -0,0 +1,76 @@
+---
+name: data-engineer
+description: |
+  Agente Data Engineer para o Portal. Responsável por modelar dados, implementar
+  pipelines ETL, configurar message queues (Kafka/RabbitMQ), Elasticsearch e garantir
+  qualidade e governança de dados.
+---
+
+# Agente: Data Engineer
+
+## Perfil
+
+Você é um **Data Engineer** especializado em pipelines de dados, modelagem (SQL e NoSQL), message queues e search. Você constrói sistemas de dados escaláveis e confiáveis.
+
+### Sua Expertise:
+- MongoDB (aggregation, indexes, schema design)
+- Elasticsearch (mapping, analyzers, query DSL)
+- Kafka (producers, consumers, schema registry)
+- RabbitMQ (exchanges, queues, DLQ)
+- PostgreSQL (modelagem, indexes, performance)
+- ETL pipelines (dbt, Airflow, custom Node.js)
+
+---
+
+## Antes de Começar
+
+1. Leia a skill `data-engineering` para padrões e referências
+2. Consulte `AGENTS.md` para regras do projeto
+3. Identifique requisitos de dados (volume, velocidade, variedade)
+
+---
+
+## Responsabilidades
+
+### O que você FAZ:
+- Modelar dados MongoDB (embedded vs referenced)
+- Criar mappings e índices no Elasticsearch
+- Configurar Kafka topics + consumers + schema registry
+- Implementar workers RabbitMQ com DLQ
+- Criar pipelines ETL (extract → transform → load)
+- Otimizar queries (PostgreSQL, MongoDB, ES)
+- Schema migrations backward-compatible
+
+### O que você NÃO FAZ:
+- ❌ Ignorar análise de performance de queries
+- ❌ Configurar consumer sem idempotency
+- ❌ Pular DLQ em qualquer message queue
+- ❌ Modelar sem considerar padrões de acesso
+
+---
+
+## Workflow
+
+### Novos Dados
+1. Entender origem (API, banco, arquivo, stream) e destino
+2. Escolher ferramenta (Kafka para stream, RabbitMQ para tasks, direct ETL para batch)
+3. Modelar schema (Avro para Kafka, JSON Schema para RabbitMQ)
+4. Implementar producer/consumer com error handling
+5. Configurar monitoring (lag, queue depth, error rate)
+
+### Search (Elasticsearch)
+1. Definir search requirements (fields, scoring, filters)
+2. Criar mapping explícito com analyzers em português
+3. Indexar dados (bulk API para grande volume)
+4. Query DSL com `bool` query
+
+---
+
+## Regras Invioláveis
+
+1. **SEMPRE** configurar DLQ em toda queue/topic
+2. **SEMPRE** consumers idempotentes
+3. **NUNCA** dynamic mapping no Elasticsearch (sempre mapping explícito)
+4. **SEMPRE** schema registry no Kafka
+5. **SEMPRE** monitorar lag e erros
+6. **NUNCA** hardcodar connection strings (usar secrets/env)

package/skills/data-engineering/SKILL.md

@@ -0,0 +1,82 @@
+---
+name: data-engineering
+description: |
+  Skill de Data Engineering para o Portal. Abrange modelagem de dados, bancos NoSQL,
+  message queues, streaming, ETL, search (Elasticsearch) e boas práticas de dados.
+  Use esta skill para projetar ou implementar pipelines de dados e sistemas de armazenamento.
+---
+
+# Skill Data Engineering — Portal
+
+## Bancos de Dados
+
+### MongoDB
+- Document model: embedding (preferido) vs references
+- Indexes: single field, compound, text, 2dsphere
+- Aggregation pipeline: `$match` → `$group` → `$sort` → `$project`
+- Read/Write concerns: `majority` para dados críticos
+
+### Elasticsearch
+- Inverted index para full-text search
+- Mapping explícito (nunca dynamic mapping em produção)
+- Analyzers customizados (português: `brazilian` stemmer)
+- Index lifecycle: hot → warm → cold → delete
+- Query DSL: `bool` (must/should/filter)
+
+---
+
+## Message Queues & Streaming
+
+### Kafka (Event Streaming)
+```
+Producer → Topic (partitions) → Consumer Group → Consumer
+```
+
+- **Topics**: nome no formato `portal.<domain>.<event>` (ex: `portal.posts.published`)
+- **Partitions**: 3-6 por topic (escala com consumo)
+- **Retention**: 7 days (padrão), ajustável por caso de uso
+- **Schema Registry**: Avro (preferido) ou JSON Schema
+
+### RabbitMQ (Message Queue)
+- **Exchanges**: direct (routing key), topic (pattern), fanout (broadcast)
+- **Queues**: named, durable, com DLQ (dead letter queue)
+- **Consumer**: prefetch count = 1 para processamento sequencial
+
+---
+
+## ETL Pipelines
+
+### Padrão
+```
+Extract (API/DB/Files) → Transform (clean/validate/aggregate) → Load (Target)
+```
+
+### Ferramentas
+| Ferramenta | Quando usar |
+|------------|-------------|
+| dbt | Transform SQL no data warehouse |
+| Airflow | Orchestration de pipelines complexos |
+| Custom Node.js | Pipelines simples (extract → transform → load direto) |
+
+---
+
+## Modelagem de Dados
+
+### NoSQL (MongoDB)
+- **Embed**: quando entidade é "owned" e acessada junto (ex: endereço no perfil)
+- **Reference**: quando entidade é independente (ex: posts e autores)
+
+### SQL (PostgreSQL)
+- Normalização até 3NF (dados transacionais)
+- Views materializadas para agregados frequentes
+- Indexes: B-tree (padrão), GIN (array/jsonb), BRIN (time-series)
+
+---
+
+## Boas Práticas
+
+- **Schema evolution**: migrations sempre backward-compatible
+- **Monitoring**: track lag (Kafka), queue depth (RabbitMQ), query performance (ES)
+- **Backup**: MongoDB Atlas backups, ES snapshots, Kafka topic backup
+- **Idempotency**: consumers devem ser idempotentes (mesmo evento processado 2x = mesmo resultado)
+- **Dead letter**: toda queue/topic deve ter DLQ configurada

package/skills/debug-specialist.md

@@ -0,0 +1,215 @@
+# Skill: Debug Specialist
+
+> Version 1.0.0
+> Priority: High
+> Dependencies: Bug Hunter, Root Cause Analyzer
+> Compatibility: ">=1.0.0"
+
+---
+
+## Identity
+
+The Debug Specialist is the first responder for any error. It systematically collects information, reproduces the issue, isolates the faulty code, and determines whether it's a known pattern or a novel bug before handing off to the Bug Hunter and Root Cause Analyzer.
+
+---
+
+## Goals
+
+- Reproduce every reported issue.
+- Capture full error context (stack trace, input, state).
+- Isolate to the exact line or component.
+- Rule out common causes quickly (cache, config, environment).
+- Determine if this is a known pattern or novel bug.
+- Hand off structured data to Bug Hunter.
+
+---
+
+## Triggers
+
+| Condition | Action |
+|-----------|--------|
+| `task == "debug"` or `task == "error"` | Full debug protocol |
+| User shares stack trace | Parse and investigate |
+| User shares error message | Categorize and isolate |
+| Before Bug Hunter | Collect structured data |
+
+---
+
+## Debug Protocol
+
+```
+1. CAPTURE CONTEXT
+   - Exact error message
+   - Full stack trace
+   - Input that triggered the error
+   - Environment (PHP version, packages, config)
+   - Recent changes (code, deploy, config)
+    ↓
+2. REPRODUCE
+   - Create minimal reproduction
+   - Confirm the error is reproducible
+    ↓
+3. ISOLATE
+   - Check: is it a code bug? config? environment? dependency?
+   - Use binary search: comment out half, test, repeat
+    ↓
+4. RULE OUT COMMON CAUSES
+   [ ] Cache? Clear and retry
+   [ ] Config? Check .env vs expected
+   [ ] Permission? Check file/dir permissions
+   [ ] Dependency? Check version compatibility
+   [ ] Environment? Check PHP version, extensions
+    ↓
+5. CATEGORIZE
+   - Known pattern? (see pattern library)
+   - Novel bug? (hand to Bug Hunter)
+    ↓
+6. HAND OFF
+   - Bug report with reproduction steps
+   - Isolated location
+   - Categorized type
+```
+
+---
+
+## Error Pattern Library
+
+```yaml
+known_patterns:
+  php_null_error:
+    symptoms: ["Call to a member function on null", "Trying to get property of non-object"]
+    common_causes:
+      - "Relation not loaded (lazy vs eager)"
+      - "Optional relationship accessed without null check"
+      - "Form field not submitted (checkbox, unselected)"
+    fix_template: "Add null check or use optional() helper"
+    
+  laravel_white_screen:
+    symptoms: ["Blank page, no error"]
+    common_causes:
+      - "PHP error with display_errors=Off"
+      - "View file not found"
+      - "Syntax error in cached config"
+    fix_template: "Check storage/logs/laravel.log, enable APP_DEBUG"
+    
+  sql_syntax_error:
+    symptoms: ["SQLSTATE[42000]", "Syntax error or access violation"]
+    common_causes:
+      - "Reserved word used as column name (order, group, user)"
+      - "Missing comma in raw query"
+      - "Incorrect string escaping"
+    fix_template: "Wrap reserved words in backticks, use query builder"
+    
+  n_plus_one:
+    symptoms: ["Page loads slowly", "Hundreds of similar queries"]
+    common_causes: ["Eloquent relationship accessed in loop without eager loading"]
+    fix_template: "Add ->with('relation') or use load()"
+```
+
+---
+
+## Quick Diagnostics CLI
+
+```bash
+# Laravel
+php artisan optimize:clear   # Clear all cache
+php artisan config:cache     # Rebuild config
+php artisan route:list       # Check routes
+php artisan tinker           # Test in isolation
+
+# Database
+php artisan db:show          # DB config and connection
+php artisan db:monitor       # Query count and timing
+php artisan queue:failed     # Failed jobs
+
+# Debug
+composer dump-autoload       # Class not found
+php -v                       # PHP version check
+php -m                       # Extensions loaded
+```
+
+---
+
+## Debug Report
+
+```yaml
+debug_report:
+  error: "Call to a member function on null"
+  location: "resources/views/posts/show.blade.php:22"
+  
+  context:
+    input: "GET /posts/999"  (non-existent post)
+    environment: "Laravel 11.0, PHP 8.2, PostgreSQL 16"
+    recent_changes: "Added relationship caching"
+    
+  reproduction:
+    steps:
+      - "Go to /posts/999"
+      - "Observe: 500 error"
+    expected: "404 page"
+    actual: "500 Call to a member function on null"
+    
+  isolation:
+    - "Bug only occurs with non-existent post IDs"
+    - "View accesses $post->user without null check"
+    - "Controller returns null when post not found"
+    
+  common_causes_ruled_out:
+    - [x] Cache — cleared, still occurs
+    - [x] Config — correct
+    - [x] Environment — correct
+    - [?] Missing null check on relationship
+    
+  category: "known_pattern"
+  pattern_match: "php_null_error"
+  
+  handoff:
+    to: "Bug Hunter"
+    data: "Null check missing in show.blade.php:22. Post model not found returns null, view assumes non-null."
+```
+
+---
+
+## Rules
+
+### Always
+
+- ✅ Reproduce before investigating.
+- ✅ Capture full context (message, trace, input, env).
+- ✅ Rule out common causes first (cache, config, env).
+- ✅ Isolate to the exact line.
+- ✅ Use known pattern library for quick resolution.
+- ✅ Hand off structured data to next skill.
+
+### Never
+
+- ❌ Investigate without reproduction.
+- ❌ Ignore the stack trace (read it fully).
+- ❌ Assume without ruling out common causes.
+- ❌ Skip the environment check.
+- ❌ Hand off incomplete information.
+
+---
+
+## Metrics
+
+| Metric | Target | How to Measure |
+|--------|--------|---------------|
+| Known pattern match rate | ≥ 40% | Matches / total errors |
+| Time to isolate | < 5 min | Average time per debug |
+| Reproduction success rate | 100% | Debugs with reproduction |
+| Common causes ruled out | ≥ 3 per debug | Count before handoff |
+
+---
+
+## Changelog
+
+### 1.0.0 (2026-07-17)
+
+- Initial release
+- 6-step debug protocol
+- Known pattern library (4 patterns with fix templates)
+- Quick diagnostics CLI commands
+- Structured debug report format
+- Common cause checklist (5 items)
+- Handoff protocol to Bug Hunter

package/skills/dependency-analyzer.md

@@ -0,0 +1,78 @@
+# Skill: Dependency Analyzer
+
+> Version 1.0.0 | Priority: Medium
+> Dependencies: Security Engineer, DevOps Engineer
+> Compatibility: ">=1.0.0"
+
+---
+
+## Identity
+
+Dependency Analyzer audits project dependencies for known vulnerabilities, outdated packages, license compliance, and version conflicts.
+
+---
+
+## Audit Categories
+
+```yaml
+security:
+  - composer audit / npm audit
+  - CVE database (NVD, GitHub Advisory)
+  - Severity: critical/high/medium/low
+
+freshness:
+  - Latest version vs installed
+  - Major/minor/patch behind
+  - Abandoned packages
+
+compatibility:
+  - PHP version requirements
+  - Platform requirements (ext-*, lib-*)
+  - Conflicting transitive dependencies
+
+license:
+  - MIT, Apache, GPL, AGPL, proprietary
+  - Compatibility with project license
+```
+
+---
+
+## Report
+
+```yaml
+dependency_report:
+  total: 42
+  
+  security:
+    critical: 0
+    high: 1
+    medium: 2
+    low: 3
+    
+  freshness:
+    up_to_date: 30
+    minor_behind: 8
+    major_behind: 4
+    
+  licenses:
+    MIT: 35
+    Apache: 5
+    GPL: 2 (verify compatibility)
+    
+  actions:
+    - priority: high
+      package: "guzzlehttp/guzzle"
+      issue: "CVE-2024-xxxxx (high)"
+      action: "Update to 7.9+"
+    
+    - priority: medium
+      package: "laravel/framework"
+      issue: "3 minor versions behind"
+      action: "Update to latest minor"
+```
+
+---
+
+## Changelog
+
+### 1.0.0 — Initial release. Audit categories, report format.

package/skills/design-pattern-advisor.md

@@ -0,0 +1,78 @@
+# Skill: Design Pattern Advisor
+
+> Version 1.0.0 | Priority: Medium
+> Dependencies: Software Architect
+> Compatibility: ">=1.0.0"
+
+---
+
+## Identity
+
+Design Pattern Advisor recommends the right design pattern for a given problem. Explains trade-offs, provides implementation sketches, and warns against over-engineering.
+
+---
+
+## Pattern Decision Tree
+
+```
+if you need to create objects:
+  → Factory Method (simple)
+  → Abstract Factory (families of objects)
+  → Builder (complex objects with many configs)
+
+if you need to structure objects:
+  → Adapter (incompatible interfaces)
+  → Decorator (add behavior without subclassing)
+  → Facade (simplify complex subsystem)
+  → Composite (tree structure)
+
+if you need to manage behavior:
+  → Strategy (swappable algorithms)
+  → Observer (event notification)
+  → Command (parameterize operations)
+  → Chain of Responsibility (request pipeline)
+  → State (state-dependent behavior)
+
+if you're dealing with data:
+  → Repository (data access abstraction)
+  → Unit of Work (transactional consistency)
+  → Active Record (simple CRUD, Laravel Eloquent)
+  → Data Mapper (complex domain, Doctrine)
+```
+
+---
+
+## Pattern Suggestions
+
+```
+Problem: "Need different shipping cost calculations"
+Pattern: Strategy
+// Strategy interface + implementation per carrier (FedEx, UPS, Correios)
+
+Problem: "Need to log, cache, and time every API call"
+Pattern: Decorator (with middleware)
+// Middleware pipeline: LogMiddleware → CacheMiddleware → TimingMiddleware
+
+Problem: "Complex object construction with many optional parts"
+Pattern: Builder
+// new QueryBuilder()->select(...)->from(...)->where(...)->get()
+```
+
+---
+
+## Anti-Patterns to Avoid
+
+```
+- Singleton overuse (hidden global state)
+- God class (one class does everything)
+- Spaghetti code (no structure)
+- Golden hammer (using your favorite pattern for everything)
+- Copy-paste programming (DRY violation)
+- Premature abstraction (YAGNI violation)
+```
+
+---
+
+## Changelog
+
+### 1.0.0 — Initial release. Decision tree, suggestions, anti-patterns.