ix-docs

Generate onboarding-ready architectural documentation by applying graph-first analysis to produce narrative-first explanations of system mechanics and navigation paths. This approach highlights critical modules while suppressing low-value verbosity, ensuring new engineers understand the codebase structure quickly.

Documenting system architecture for LLM reasoning requires narrative-first documentation with a selective reference layer. By applying importance-weighted graph analysis, it provides compact summaries of key modules, enabling both human engineers and LLMs to navigate complex codebases effectively.

Split documentation output for large repositories by using the --split flag to enable auto-splitting. This flexible output format divides the narrative-first architectural documentation and selective reference layers into manageable, navigable sections without losing structural context.

Control dependency analysis depth in generated docs using configurable output flags like --full and --style. The --full flag expands the selective reference layer to provide deeper coverage of key modules, classes, and services, while --style adjusts the narrative format.

Narrative-first documentation works with rare code reads by relying on graph-first analysis and strict limits on low-value verbosity. It prioritizes importance-weighted selective references over exhaustive codebase scanning, explaining why the system exists and how it works efficiently.

Avoid auto-generated documentation when a codebase lacks clear architectural boundaries or stable module structures, as graph-first analysis relies on identifiable dependencies. Without coherent system modeling, the importance-weighted narrative may fail to provide accurate onboarding navigation guidance.