map-codebase

Map the codebase by deep-reading every file to generate skeleton LLDs, a bottom-up HLD, and EARS specs. This process also creates tree-mirrored arrow docs to help you flesh out the design skeletons.

Reverse-engineer design by deep-reading the declared scope and proposing lens-based clustering options. This generates an overlay-ready docs/index scaffold and arrow docs that prompt you to flesh out the recovered architecture.

Generate EARS specs and skeleton LLDs by invoking the mapping workflow against your live repository. It deeply reads every file to ground the design with concrete evidence and outputs the specifications bottom-up per segment.

Yes, codebase mapping supports scoped or whole-project operation. It requires repository access and uses explicit STOP gates at each phase, allowing you to control whether you map a specific directory or the entire project.

The mapping proposes 3–5 fundamentally different clustering lenses, not just variants. It requires deep-reading every file in scope, which may be time-consuming for large repositories, and it only writes docs to specific directories.

Yes, mapping a codebase requires access to repository contents to deep-read every file. It also needs the ability to write generated skeleton LLDs, HLD, EARS, and arrow docs under the docs/arrows, docs/intent, and docs/high-level-design paths.