DRY might be wrong for agent documentation

This article challenges the DRY (Don't Repeat Yourself) principle in the context of documentation written for AI agents. It argues that while reducing duplication simplifies maintenance, it forces readers (especially agents) to assemble fragmented information from multiple sources, increasing retrieval costs and inconsistency risks. The author suggests optimizing for complete workflows by providing all necessary instructions in one resource, even if it means duplicating content, while still using progressive disclosure to manage context. This shift in perspective is relevant for technical writers and developers building agent-friendly documentation.