Describes a reusable “prompt pattern” framework plus a catalog of patterns (e.g., output templates, question decomposition, constraints, and interaction patterns) for systematically steering LLM behavior. For Process Safety Management (PSM), these patterns are useful where outputs must be consistent, auditable, and aligned to procedures—such as generating structured PHA/LOPA checklists, drafting MOC review questions, creating incident-investigation summaries with required fields, or producing training Q&A aligned to site standards. The pattern approach helps teams standardize prompts across users and shifts, reduce variability, and build repeatable prompt “recipes” that can be reviewed and improved like other safety-critical work products.
