Security engineers and AI engineers responsible for the security posture of LLM-powered applications need to systematically test for prompt injection, indirect prompt injection via retrieved documents, system prompt extraction, and model DoS vulnerabilities before production launch or security audits.