izanagi-ai 2.0.0

package/skills/INDEX.md

@@ -0,0 +1,434 @@
+# Skills Index
+
+> Complete registry of all skills in the Nexus AI framework.
+> Updated: 2026-07-22 | Total: 137 skills
+
+---
+
+## Core Skills (8)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Decision Engine | ✅ v1.0.0 | `core/decision-engine.md` |
+| Context Engine | ✅ v1.0.0 | `core/context-engine.md` |
+| Token Manager | ✅ v1.0.0 | `core/token-manager.md` |
+| Compression Engine | ✅ v1.0.0 | `core/compression-engine.md` |
+| Reflection Engine | ✅ v1.0.0 | `core/reflection-engine.md` |
+| Evolution Engine | ✅ v1.0.0 | `core/evolution-engine.md` |
+| Quality Gates | ✅ v1.0.0 | `core/quality-gates.md` |
+| Planning Engine | ✅ v1.0.0 | `core/planning-engine.md` |
+
+---
+
+## Architecture Skills (10)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Software Architect | ✅ v1.0.0 | `skills/software-architect.md` |
+| Clean Architecture | ✅ v1.0.0 | `architecture/clean-architecture.md` |
+| Hexagonal Architecture | ✅ v1.0.0 | `architecture/hexagonal-architecture.md` |
+| CQRS Specialist | ✅ v1.0.0 | `architecture/cqrs-specialist.md` |
+| Event-Driven Architect | ✅ v1.0.0 | `architecture/event-driven-architect.md` |
+| DDD Specialist | ✅ v1.0.0 | `architecture/ddd-specialist.md` |
+| Microservices Expert | ✅ v1.0.0 | `architecture/microservices-expert.md` |
+| Monolith Expert | ✅ v1.0.0 | `architecture/monolith-expert.md` |
+| Repository Pattern | ✅ v1.0.0 | `architecture/repository-pattern.md` |
+| Unit of Work | ✅ v1.0.0 | `architecture/unit-of-work.md` |
+
+---
+
+## Coding Skills (13)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Backend Engineer | ✅ v1.0.0 | `coding/backend-engineer.md` |
+| Frontend Engineer | ✅ v1.0.0 | `coding/frontend-engineer.md` |
+| API Designer | ✅ v1.0.0 | `coding/api-designer.md` |
+| Laravel Specialist | ✅ v1.0.0 | `coding/laravel-specialist.md` |
+| PHP Specialist | ✅ v1.0.0 | `coding/php-specialist.md` |
+| JavaScript Specialist | ✅ v1.0.0 | `coding/javascript-specialist.md` |
+| TypeScript Specialist | ✅ v1.0.0 | `coding/typescript-specialist.md` |
+| Python Specialist | ✅ v1.0.0 | `coding/python-specialist.md` |
+| React Specialist | ✅ v1.0.0 | `coding/react-specialist.md` |
+| Vue Specialist | ✅ v1.0.0 | `coding/vue-specialist.md` |
+| Node.js Specialist | ✅ v1.0.0 | `coding/nodejs-specialist.md` |
+| Java Specialist | ✅ v1.0.0 | `coding/java-specialist.md` |
+| C# Specialist | ✅ v1.0.0 | `coding/csharp-specialist.md` |
+
+---
+
+## Security Skills (3)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Security Engineer | ✅ v1.0.0 | `security/security-engineer.md` |
+| OWASP Auditor | ✅ v1.0.0 | `security/owasp-auditor.md` |
+| Pentest Reviewer | ✅ v1.0.0 | `security/pentest-reviewer.md` |
+
+---
+
+## Quality Skills (12)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Senior Code Reviewer | ✅ v1.0.0 | `skills/senior-code-reviewer.md` |
+| Clean Code Validator | ✅ v1.0.0 | `skills/clean-code-validator.md` |
+| SOLID Validator | ✅ v1.0.0 | `skills/solid-validator.md` |
+| DRY/KISS/YAGNI Validator | ✅ v1.0.0 | `skills/dry-kiss-yagni-validator.md` |
+| Complexity Analyzer | ✅ v1.0.0 | `skills/complexity-analyzer.md` |
+| Bug Prevention | ✅ v1.0.0 | `skills/bug-prevention.md` |
+| Design Pattern Advisor | ✅ v1.0.0 | `skills/design-pattern-advisor.md` |
+| Refactoring Specialist | ✅ v1.0.0 | `skills/refactoring-specialist.md` |
+| Technical Debt Analyzer | ✅ v1.0.0 | `skills/technical-debt-analyzer.md` |
+| Breaking Change Detector | ✅ v1.0.0 | `skills/breaking-change-detector.md` |
+| Performance Optimizer | ✅ v1.0.0 | `skills/performance-optimizer.md` |
+| Scalability Expert | ✅ v1.0.0 | `skills/scalability-expert.md` |
+
+---
+
+## Testing Skills (4)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Unit Test Engineer | ✅ v1.0.0 | `testing/unit-test-engineer.md` |
+| Integration Test Engineer | ✅ v1.0.0 | `testing/integration-test-engineer.md` |
+| E2E Test Engineer | ✅ v1.0.0 | `testing/e2e-test-engineer.md` |
+| Mocking Specialist | ✅ v1.0.0 | `testing/mocking-specialist.md` |
+
+---
+
+## Database Skills (6)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Database Engineer | ✅ v1.0.0 | `database/database-engineer.md` |
+| SQL Optimizer | ✅ v1.0.0 | `database/sql-optimizer.md` |
+| PostgreSQL Specialist | ✅ v1.0.0 | `database/postgresql-specialist.md` |
+| MySQL Specialist | ✅ v1.0.0 | `database/mysql-specialist.md` |
+| SQL Server Specialist | ✅ v1.0.0 | `database/sqlserver-specialist.md` |
+| Redis Specialist | ✅ v1.0.0 | `database/redis-specialist.md` |
+
+---
+
+## DevOps Skills (8)
+
+| Skill | Status | File |
+|-------|--------|------|
+| DevOps Engineer | ✅ v1.0.0 | `devops/devops-engineer.md` |
+| Docker Expert | ✅ v1.0.0 | `devops/docker-expert.md` |
+| Kubernetes Specialist | ✅ v1.0.0 | `devops/kubernetes-specialist.md` |
+| Git Expert | ✅ v1.0.0 | `devops/git-expert.md` |
+| Git Flow Specialist | ✅ v1.0.0 | `devops/git-flow-specialist.md` |
+| CI/CD Specialist | ✅ v1.0.0 | `devops/ci-cd-specialist.md` |
+| Linux Specialist | ✅ v1.0.0 | `devops/linux-specialist.md` |
+| Windows Specialist | ✅ v1.0.0 | `devops/windows-specialist.md` |
+
+---
+
+## Engineering Roles (4)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Principal Engineer | ✅ v1.0.0 | `skills/principal-engineer.md` |
+| Staff Engineer | ✅ v1.0.0 | `skills/staff-engineer.md` |
+| Tech Lead | ✅ v1.0.0 | `skills/tech-lead.md` |
+| CTO Advisor | ✅ v1.0.0 | `skills/cto-advisor.md` |
+
+---
+
+## Debugging & Root Cause (3)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Bug Hunter | ✅ v1.0.0 | `skills/bug-hunter.md` |
+| Debug Specialist | ✅ v1.0.0 | `skills/debug-specialist.md` |
+| Root Cause Analyzer | ✅ v1.0.0 | `skills/root-cause-analyzer.md` |
+
+---
+
+## Teaching Skills (6)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Professor Mode | ✅ v1.0.0 | `teaching/professor-mode.md` |
+| Mentor Mode | ✅ v1.0.0 | `teaching/mentor-mode.md` |
+| Code Explainer | ✅ v1.0.0 | `teaching/code-explainer.md` |
+| Interactive Teaching | ✅ v1.0.0 | `teaching/interactive-teaching.md` |
+| Adaptive Teaching | ✅ v1.0.0 | `teaching/adaptive-teaching.md` |
+| Learning Tracker | ✅ v1.0.0 | `teaching/learning-tracker.md` |
+
+---
+
+## Memory Skills (6)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Memory Manager | ✅ v1.0.0 | `memory/memory-manager.md` |
+| Session Compression | ✅ v1.0.0 | `memory/session-compression.md` |
+| Conversation Summarizer | ✅ v1.0.0 | `memory/conversation-summarizer.md` |
+| Context Recovery | ✅ v1.0.0 | `memory/context-recovery.md` |
+| Smart Recall | ✅ v1.0.0 | `memory/smart-recall.md` |
+| Long-Term Project Memory | ✅ v1.0.0 | `memory/long-term-project-memory.md` |
+
+---
+
+## Optimization Skills (3)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Token Reducer | ✅ v1.0.0 | `optimization/token-reducer.md` |
+| Prompt Optimizer | ✅ v1.0.0 | `optimization/prompt-optimizer.md` |
+| Cost Optimizer | ✅ v1.0.0 | `optimization/cost-optimizer.md` |
+
+---
+
+## Analysis Skills (5)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Requirement Analyzer | ✅ v1.0.0 | `skills/requirement-analyzer.md` |
+| Risk Analyzer | ✅ v1.0.0 | `skills/risk-analyzer.md` |
+| Dependency Analyzer | ✅ v1.0.0 | `skills/dependency-analyzer.md` |
+| Trade-off Analyzer | ✅ v1.0.0 | `skills/tradeoff-analyzer.md` |
+| Alternative Solution Generator | ✅ v1.0.0 | `skills/alternative-solution-generator.md` |
+
+---
+
+## Documentation Skills (6)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Documentation Writer | ✅ v1.0.0 | `skills/documentation-writer.md` |
+| Technical Writer | ✅ v1.0.0 | `skills/technical-writer.md` |
+| README Generator | ✅ v1.0.0 | `skills/readme-generator.md` |
+| UML Generator | ✅ v1.0.0 | `skills/uml-generator.md` |
+| Sequence Diagram Builder | ✅ v1.0.0 | `skills/sequence-diagram-builder.md` |
+| ER Diagram Builder | ✅ v1.0.0 | `skills/er-diagram-builder.md` |
+
+---
+
+## Project Management (3)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Project Manager | ✅ v1.0.0 | `skills/project-manager.md` |
+| Task Planner | ✅ v1.0.0 | `skills/task-planner.md` |
+| Release Planner | ✅ v1.0.0 | `skills/release-planner.md` |
+
+---
+
+## UX/Observability (5)
+
+| Skill | Status | File |
+|-------|--------|------|
+| UX Reviewer | ✅ v1.0.0 | `skills/ux-reviewer.md` |
+| Accessibility Reviewer | ✅ v1.0.0 | `skills/accessibility-reviewer.md` |
+| Logging Expert | ✅ v1.0.0 | `skills/logging-expert.md` |
+| Observability Expert | ✅ v1.0.0 | `skills/observability-expert.md` |
+| Monitoring Specialist | ✅ v1.0.0 | `skills/monitoring-specialist.md` |
+
+---
+
+## Self-Improvement Skills (6)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Self Correction | ✅ v1.0.0 | `skills/self-correction.md` |
+| Self Critique | ✅ v1.0.0 | `skills/self-critique.md` |
+| Continuous Improvement | ✅ v1.0.0 | `skills/continuous-improvement.md` |
+| Hallucination Detection | ✅ v1.0.0 | `skills/hallucination-detection.md` |
+| Confidence Estimator | ✅ v1.0.0 | `skills/confidence-estimator.md` |
+| Continuous Learning Engine | ✅ v1.0.0 | `skills/continuous-learning-engine.md` |
+
+
+---
+
+## AI & Agent Skills (2)
+
+| Skill | Status | File |
+|-------|--------|------|
+| AI Agent | ? v1.0.0 | `skills/ai-agent/SKILL.md` |
+| AI Agent Dev | ? v1.0.0 | `skills/ai-agent-dev/SKILL.md` |
+
+---
+
+## Architecture Patterns (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Architecture Patterns | ? v1.0.0 | `skills/architecture-patterns/SKILL.md` |
+
+---
+
+## Chaos Engineering (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Chaos Engineering | ? v1.0.0 | `skills/chaos-engineering/SKILL.md` |
+
+---
+
+## Cloud & Infrastructure (5)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Cloud Architect | ? v1.0.0 | `skills/cloud-architect/SKILL.md` |
+| Cloud Infrastructure | ? v1.0.0 | `skills/cloud-infra/SKILL.md` |
+| IaC & Terraform | ? v1.0.0 | `skills/iac-terraform/SKILL.md` |
+| Serverless & Edge | ? v1.0.0 | `skills/serverless-edge/SKILL.md` |
+| SRE & Reliability | ? v1.0.0 | `skills/sre-reliability/SKILL.md` |
+
+---
+
+## Feature Flags (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Feature Flags | ? v1.0.0 | `skills/feature-flags/SKILL.md` |
+
+---
+
+## i18n & Localization (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| i18n & Localization | ? v1.0.0 | `skills/i18n-l10n/SKILL.md` |
+
+---
+
+## Legacy Migration (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Legacy Migration | ? v1.0.0 | `skills/legacy-migration/SKILL.md` |
+
+---
+
+## Mobile Development (2)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Mobile Development | ? v1.0.0 | `skills/mobile-dev/SKILL.md` |
+| Mobile Engineer | ? v1.0.0 | `skills/mobile-engineer/SKILL.md` |
+
+---
+
+## Privacy (2)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Privacy Engineer | ? v1.0.0 | `skills/privacy-engineer/SKILL.md` |
+| Security & Privacy | ? v1.0.0 | `skills/security-privacy/SKILL.md` |
+
+---
+
+## Quality Assurance (2)
+
+| Skill | Status | File |
+|-------|--------|------|
+| QA (Checklist) | ? v1.0.0 | `skills/qa/SKILL.md` |
+| QA Engineer | ? v1.0.0 | `skills/qa-engineer/SKILL.md` |
+
+---
+
+## Token Economics (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Economia de Tokens | ? v1.0.0 | `skills/economia-tokens/SKILL.md` |
+
+---
+
+## Session Handoff (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Handoff de Sess�o | ? v1.0.0 | `skills/handoff-sessao/SKILL.md` |
+
+---
+
+## Project Memory (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Mem�ria de Projeto | ? v1.0.0 | `skills/memoria-projeto/SKILL.md` |
+
+---
+
+## Professor Mode (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Modo Professor | ? v1.0.0 | `skills/professor-modo/SKILL.md` |
+
+---
+
+## WebAssembly (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| WebAssembly | ? v1.0.0 | `skills/wasm/SKILL.md` |
+
+---
+
+## Web Performance & SEO (2)
+
+| Skill | Status | File |
+|-------|--------|------|
+| Web Performance Engineer | ? v1.0.0 | `skills/web-perf-engineer/SKILL.md` |
+| Web Performance & SEO | ? v1.0.0 | `skills/web-perf-seo/SKILL.md` |
+
+---
+
+## WebSocket & Real-Time (1)
+
+| Skill | Status | File |
+|-------|--------|------|
+| WebSocket & Real-Time | ? v1.0.0 | `skills/websocket-realtime/SKILL.md` |
+
+---
+
+## Stats
+
+| Metric | Value |
+|--------|-------|
+| **Total skills** | **137** |
+| Core modules | 8 |
+| Architecture | 10 |
+| Coding | 13 |
+| Security | 3 |
+| Quality | 12 |
+| Testing | 4 |
+| Database | 6 |
+| DevOps | 8 |
+| Engineering Roles | 4 |
+| Debugging | 3 |
+| Teaching | 6 |
+| Memory | 6 |
+| Optimization | 3 |
+| Analysis | 5 |
+| Documentation | 6 |
+| Project Management | 3 |
+| UX/Observability | 5 |
+| Self-Improvement | 6 |
+| AI & Agent Skills | 2 |
+| Architecture Patterns | 1 |
+| Chaos Engineering | 1 |
+| Cloud & Infrastructure | 5 |
+| Feature Flags | 1 |
+| i18n & Localization | 1 |
+| Legacy Migration | 1 |
+| Mobile Development | 2 |
+| Privacy | 2 |
+| Quality Assurance | 2 |
+| Token Economics | 1 |
+| Session Handoff | 1 |
+| Project Memory | 1 |
+| Professor Mode | 1 |
+| WebAssembly | 1 |
+| Web Performance & SEO | 2 |
+| WebSocket & Real-Time | 1 |
+
+---
+
+> "A framework is only as good as its skills."

package/skills/accessibility-reviewer.md

@@ -0,0 +1,63 @@
+# Skill: Accessibility Reviewer
+
+> Version 1.0.0 | Priority: Medium
+> Dependencies: Frontend Engineer
+> Compatibility: ">=1.0.0"
+
+---
+
+## Identity
+
+Accessibility Reviewer audits interfaces against WCAG 2.2 AA standards. Covers perceivable, operable, understandable, and robust principles.
+
+---
+
+## WCAG Principles
+
+```yaml
+perceivable:
+  - All images have alt text
+  - Captions for video content
+  - Color not sole means of conveying info
+  - Contrast ratio ≥ 4.5:1 (AA) or 3:1 (large text)
+  
+operable:
+  - All functions available via keyboard
+  - No keyboard traps
+  - Skip navigation link
+  - Touch targets ≥ 44x44px
+  
+understandable:
+  - Page language defined (lang attribute)
+  - Consistent navigation
+  - Error suggestions provided
+  - Input labels/purpose clear
+  
+robust:
+  - Semantic HTML (landmarks, headings)
+  - ARIA attributes valid and correct
+  - Works with screen readers (NVDA, VoiceOver)
+```
+
+---
+
+## Audit Tools
+
+```yaml
+automated:
+  - axe-core (browser extension)
+  - Lighthouse accessibility score
+  - WAVE evaluation tool
+
+manual:
+  - Keyboard navigation (Tab through all interactive elements)
+  - Screen reader (NVDA: Windows, VoiceOver: Mac)
+  - Zoom to 200% (no content loss)
+  - Reduced motion (prefers-reduced-motion)
+```
+
+---
+
+## Changelog
+
+### 1.0.0 — Initial release. WCAG principles, audit tools.

package/skills/agentic-coding.md

@@ -0,0 +1,56 @@
+# Skill: Agentic Coding
+
+**Versão:** 1.0.0  
+**Domínio:** Autonomous LLM Systems & Agent Workflows  
+**Budget de Tokens:** ~2500  
+
+---
+
+## Contexto & Objetivo
+Esta skill orienta a construção, refatoração e otimização de agentes autônomos de código (ex: Antigravity, OpenCode, Claude Code, Cursor Agents). Garante ciclo fechado de planejamento, execução, autoverificação e tratamento resiliente de erros.
+
+---
+
+## Princípios da Skill
+1. **Loop Autônomo Seguro**: Planejar -> Fazer -> Testar -> Refletir -> Corrigir.
+2. **Navegação Não-Destrutiva**: Nunca alterar arquivos críticos sem backup ou confirmação prévia quando incerto.
+3. **Verificação Empírica**: Nenhuma tarefa é declarada concluída sem evidência de log ou teste executado com sucesso.
+4. **Contexto Mínimo Viável**: Manter o uso de tokens eficiente com leituras focadas e resumos sintéticos.
+
+---
+
+## Workflow
+```
+[Receber Tarefa] 
+       │
+       ▼
+[Análise de Requisitos & Estudo do Codebase] 
+       │
+       ▼
+[Elaborar Plano de Execução Demarcado] 
+       │
+       ▼
+[Executar Alterações de Forma Incremental]
+       │
+       ▼
+[Executar Testes / Linters / Build] 
+       │
+       ├─► (Falha) ──► [Diagnóstico por Log & Self-Fix] ──┐
+       │                                                   │
+       └─► (Sucesso) ──► [Reflexão Final & Walkthrough] ◄──┘
+```
+
+---
+
+## Checklist de Qualidade
+- [ ] O plano explicita os arquivos que serão modificados/criados.
+- [ ] Nenhuma premissa sobre o código foi assumida sem inspeção direta.
+- [ ] Erros de execução foram analisados via logs completos.
+- [ ] Alterações foram verificadas empiricamente (testes executados).
+- [ ] Documentação e walkthrough atualizados após a execução.
+
+---
+
+## Regras
+- **Sempre**: Exigir evidência de execução antes de finalizar uma tarefa.
+- **Nunca**: Engolir exceções ou desabilitar testes quebrados para fingir sucesso.

package/skills/ai-agent/SKILL.md

@@ -0,0 +1,107 @@
+---
+name: ai-agent
+description: |
+  Skill de AI/Agent Development para o Portal. Contém padrões de integração com LLMs,
+  RAG, vector databases, LangGraph, MCP, multi-agent systems e prompt engineering.
+  Use esta skill para implementar ou revisar funcionalidades baseadas em IA generativa.
+---
+
+# Skill AI/Agent Development — Portal
+
+## Arquiteturas de IA
+
+### RAG (Retrieval-Augmented Generation)
+```
+User Query → Embedding → Vector Search (similarity) → Context + Prompt → LLM Response
+```
+
+- **Embedding**: `text-embedding-3-small` / `text-embedding-ada-002`
+- **Vector DB**: Supabase pgvector, Pinecone, Qdrant, Chroma
+- **Chunking**: RecursiveCharacterTextSplitter (500-1000 tokens, 10% overlap)
+- **Retrieval**: similarity search (cosine distance), hybrid search (keyword + vector)
+
+### Agentes Autônomos
+```
+Tool 1 ─┐
+Tool 2 ──┤  LLM (ReAct loop) → Action → Observation → Next Action
+Tool 3 ─┘
+```
+
+- **Framework**: LangGraph (preferido), CrewAI, AutoGen, Vercel AI SDK
+- **Pattern**: ReAct (Reasoning + Acting), Plan-and-Execute, Reflection
+- **Memory**: BufferWindowMemory, SummarizeMemory, VectorStoreMemory
+
+### MCP (Model Context Protocol)
+```
+Host (app) → MCP Client → MCP Server → Tool/Resource/Context
+```
+
+- Usar MCP para expor ferramentas internas ao LLM
+- Cada MCP Server = 1 domínio (banco, API externa, sistema interno)
+
+---
+
+## Stack Recomendada (2026)
+
+| Componente | Sugestão |
+|------------|----------|
+| LLM API | OpenAI GPT-4o, Claude 4, Gemini 2.0 |
+| Vector DB | Supabase pgvector (preferido) |
+| Agent Framework | LangGraph |
+| AI SDK | Vercel AI SDK (`ai`) |
+| Embeddings | `openai` ou `cohere` |
+| MCP | `@modelcontextprotocol/sdk` |
+
+---
+
+## Prompt Engineering
+
+### Estrutura de System Prompt
+```
+Role: [defina o papel do assistente]
+Context: [contexto relevante]
+Rules: [regras e constraints]
+Tools: [ferramentas disponíveis]
+Output Format: [formato esperado da resposta]
+```
+
+### Anti-Padrões
+- ❌ Prompts genéricos sem contexto específico do projeto
+- ❌ Instruções contraditórias (ex: "seja conciso" + "explique detalhadamente")
+- ❌ Prompt injection sem sanitização de input do usuário
+- ❌ Expor system prompt em respostas ao usuário
+
+---
+
+## Implementação no Projeto
+
+### Vercel AI SDK
+```tsx
+import { generateText, streamText } from "ai";
+import { openai } from "@ai-sdk/openai";
+
+const { text } = await generateText({
+  model: openai("gpt-4o"),
+  system: "Você é um assistente do Portal...",
+  prompt: userQuery,
+});
+```
+
+### RAG com Supabase pgvector
+```tsx
+const { data: documents } = await supabase.rpc("match_documents", {
+  query_embedding: embedding,
+  match_threshold: 0.78,
+  match_count: 5,
+});
+```
+
+---
+
+## Segurança em IA
+
+- **Guardrails**: validar output do LLM antes de exibir ao usuário
+- **Rate limiting**: limitar chamadas por usuário/IP
+- **Content filtering**: bloquear prompts maliciosos (injection, jailbreak)
+- **Data privacy**: nunca enviar PII (CPF, RG) como contexto para o LLM
+- **Audit logging**: logar todas as interações IA (anonimizadas)