PromptFuzz systematically discovers vulnerabilities in LLM-powered applications through automated adversarial testing. It generates, mutates, and executes attack prompts against any LLM endpoint, then reports findings with severity ratings, reproduction steps, and CI integration.