What problem does it solve?
Initializes an AI-friendly, navigable documentation layer so agents can quickly understand what exists in each directory and how to work there correctly.
Core Features & Use Cases
- Hierarchical AGENTS.md Generation: Produces AGENTS.md files across the codebase and links them via parent reference tags for an explicit documentation tree.
- Directory Mapping and Level-by-Level Planning: Scans directories recursively, plans work by depth, and generates parent-first to keep references valid.
- Update and Preservation of Manual Notes: When AGENTS.md already exists, updates auto-generated sections while preserving content marked as manual.
- Validation for Completeness and Parent Resolution: Checks that parent references resolve and that no orphaned or missing AGENTS.md files remain.
Quick Start
Ask the AI to run the deepinit skill to generate or update hierarchical AGENTS.md files across the repository, preserving any manual sections.