reversa-writer

Executable specifications are generated from extracted legacy knowledge by creating unit-scoped folders containing canonical requirements.md, design.md, and tasks.md files. This process reads extracted knowledge artifacts and converts them into operational, traceable specifications.

Reverse documentation turns extracted legacy system knowledge into operational specifications so AI coding agents can safely evolve existing software. You need it when you must modernize legacy systems but lack original specifications for safe modifications.

Unit-scoped specifications are created by organizing output into module, endpoint, use-case, feature, hybrid, or custom granularity folders. The process reads surface.json modules and features to enumerate units and generates aligned requirements, design, and tasks files.

Prior context files are required before generating documentation because the process reads .reversa/state.json and .reversa/config.toml for configuration. It must run after Reversa has generated prior context and extracted knowledge to produce accurate specifications.

Generating all legacy system specifications at once is intentionally prevented by controlled scope guardrails. The workflow follows a never generate everything at once approach, checking Reversa state and config to process only selected units at a time.