Cria prompts de geração de imagem pelo método 3W1H de Mário Lúcio (Who/What/Where/How) — a partir de uma descrição abstrata via perguntas 3W+H1, ou fazendo engenharia reversa de um arquivo de imagem. Entrega prompt pt-br + prompt en + alt text em blocos separados. USE WHEN criar prompt de imagem, prompt 3W1H, 3WH1, método 3w1h, gerar prompt para IA de imagem, descrever imagem em prompt, engenharia reversa de imagem, prompt Midjourney/DALL-E/Flux estruturado. NOT FOR gerar a imagem em si (esta skill produz apenas o prompt).
Fundamental Prompt Engineering skill. Use when the user asks to create, improve, or refine prompts, define personas, use delimiters, provide examples (few-shot), apply control levels (complexity, tone, sentiment, perspective, focus, surprise, detail, originality, abstraction), use Chain-of-Thought (CoT), Chain-of-Verification (CoVe), KD-CoT, Interactive Prompt, or reduce hallucinations with Self-Reflection. Based on the book "Prompts em Ação: Engenharia de Prompts para Leigos" by Sandeco (2024).
Specialized skill in Guardrails for LLMs. Use when the user asks to create secure prompts, implement guardrails, protect against prompt injection, validate AI outputs, apply prompt design patterns, build multi-agent debate systems with human feedback, use Rephrase and Respond (RaR), or integrate the Guardrails AI library (Python). Based on the book "Engenharia de Prompts II – Guardrails" by Sandeco.
Portable Terraform AWS project bootstrap skill. Use when the user asks to run `/aws-terraform-template` or `/aws-terraform-template init`, create a minimal AWS Terraform infrastructure repository, bootstrap Terraform for AWS, initialize an AWS IaC project, or generate agent-specific instructions for Codex, Claude Code, Kiro, or Antigravity.