Reverse Engineer

This is step 2 in a six-step reverse engineering workflow that generates 11 documentation files from your existing codebase. It reads a state file from step 1 to determine whether you're going greenfield (business logic only, tech-agnostic) or brownfield (full implementation details, tech-prescriptive), then spawns parallel analysis tasks across backend, frontend, infrastructure, testing, and business context. The output lands in docs/reverse-engineering/ and gets pinned to a git commit hash for incremental updates. It's built for monorepos and Nx projects but handles generic apps too. The execution model is solid: it validates prerequisites, retries failed subagents with fallbacks, and documents missing components instead of silently skipping them. Takes 30 to 45 minutes and requires the initial analysis step to run first.