What problem does it solve? Brownfield projects often lack documentation that AI agents and new developers can use to understand the codebase. This Skill scans an existing project and produces a complete documentation set (index, architecture, source tree, API contracts, data models, development guides) ready for brownfield PRD planning. ## Core Features & Use Cases - Three Scan Levels: Quick (pattern-based, no source reading), Deep (reads critical directories), and Exhaustive (reads every source file), with batching and write-as-you-go output to keep context lean. - Project Type Detection: Classifies projects against 12 types (web, mobile, backend, CLI, library, desktop, game, data, extension, infra, embedded) using a documentation-requirements CSV, and handles monolith, monorepo, and multi-part structures. - Resumable State Tracking: Maintains a project-scan-report.json state file so interrupted runs can resume from the exact step, archiving state files older than 24 hours. - Deep-Dive Mode: Performs exhaustive full-file analysis of a chosen folder, feature, or API group, producing dependency graphs, data flow traces, and contributor guidance. - Use Case: Point it at a legacy repository, choose a deep scan, and receive an index.md plus architecture, source-tree, and development-guide documents that an AI agent can immediately use for feature planning. ## Quick Start Ask the assistant to document this project and choose a scan level when prompted.
