What problem does it solve? Reverse-engineered knowledge about a legacy system is often scattered and unstructured, making it impossible to reimplement the system faithfully. This Skill converts extracted legacy knowledge into formal, traceable specifications organized as folder-per-unit, each containing requirements.md, design.md, and tasks.md contracts detailed enough for an AI agent to reimplement the functionality without access to the original code. ## Core Features & Use Cases - Granularity-aware unit planning: Resolves spec units from modules, endpoints, use cases, features, hybrid layouts, or custom folders based on the reversa config.toml settings. - Incremental, checkpointed generation: Produces one file at a time with user confirmation, saves progress to .reversa/state.json, and suggests preventive pauses after every 3 units to preserve quality in long sessions. - Confidence-tagged traceability: Marks every statement as confirmed, inferred, or gap, and emits global artifacts like OpenAPI specs, user stories, and a code-spec traceability matrix. - Use Case: After running the earlier Reversa analysis phases on a legacy codebase, invoke this Skill to produce a complete specification tree where each module gets requirements with MoSCoW prioritization, a technical design with interface tables, and an executable task list citing legacy source files. ## Quick Start Run the reversa-writer generation phase to turn the analyzed legacy system into per-unit specification folders with requirements, design, and tasks files.
