What problem does it solve? New or unfamiliar codebases lack structured documentation, forcing developers to manually explore directory layouts, conventions, and entry points before they can work effectively. This Skill automates the creation of hierarchical AGENTS.md knowledge base files so AI agents and developers get accurate project context immediately. ## Core Features & Use Cases - Parallel Discovery: Launches concurrent explore agents to analyze project structure, entry points, conventions, anti-patterns, build systems, and test patterns simultaneously. - Complexity Scoring: Scores each directory on file count, code ratio, module boundaries, and symbol density to decide which subdirectories warrant their own AGENTS.md file. - Hierarchical Generation: Creates a root AGENTS.md plus focused subdirectory files, then deduplicates and trims content to enforce quality gates. - Use Case: After cloning a large monorepo, run the skill to produce a root knowledge base plus per-module AGENTS.md files so every future agent session starts with accurate, non-redundant project context. ## Quick Start Ask the agent to run /init-deep on your project to generate hierarchical AGENTS.md documentation for the codebase.