The Draft Init skill is the foundational entry point for Context-Driven Development, designed to bridge the gap between raw source code and AI understanding. It automates the creation of a standardized draft/ directory containing critical project context files like product definitions, tech stacks, and architectural overviews. By performing deep analysis of existing brownfield repositories and maintaining Git-synced metadata, it ensures that Claude and other AI assistants operate with a high-fidelity, up-to-date mental model of the codebase, significantly reducing hallucinations and improving implementation accuracy.
主な機能
01Git-aware metadata tracking for incremental context refreshes
02Deep brownfield codebase analysis and dependency detection
0326 GitHub stars
04Atomic file staging to prevent corrupted project configuration states
05Monorepo detection and service-level initialization logic
06Automated generation of structured documentation (Product, Tech-Stack, Architecture)
ユースケース
01Onboarding an AI assistant to a complex existing legacy codebase
02Setting up new projects with standardized Context-Driven Development patterns
03Synchronizing AI context files with recent Git commits to maintain documentation accuracy