This is a structured workflow for creating architecture documentation that AI agents can actually follow consistently. It walks you through discovery and design decisions step by step, outputting docs that prevent the usual implementation drift when multiple agents touch your codebase. The approach is collaborative rather than prescriptive, treating you as a peer who brings domain knowledge while it brings architectural structure. Uses a micro-file architecture where each step is self-contained and you control progression. Worth using when you need to lock in technical decisions before implementation starts, especially on projects where consistency across AI-generated code matters more than speed.
npx skills add https://github.com/bmad-code-org/bmad-method --skill bmad-create-architecture