Prompt engineering has become a vital skill for interacting effectively with large language models (LLMs) like ChatGPT. Prompts act as instructions, enabling rule enforcement, process automation, and tailored output. This article introduces a catalog of prompt engineering techniques, presented as reusable patterns. Inspired by software patterns, these techniques address common challenges in generating outputs and managing interactions with LLMs. Key contributions include a framework for documenting adaptable prompt structures, a catalog of proven patterns for enhancing conversations, and insights into combining patterns for complex tasks. These strategies empower users to optimize LLM performance across various domains and applications.