How does the directory scoring system work?

The skill analyzes file counts, symbol density, and import centrality to determine which directories are complex enough to require their own specific guidance file.

What is progressive disclosure in Claude Code?

It is a structural pattern where documentation is split across directories. Claude automatically loads the nearest CLAUDE.md, ensuring it only sees instructions relevant to the files it is currently editing.

Is this skill compatible with monorepos?

Yes, it is specifically designed for complex structures, allowing you to set global rules in the root and specific package rules in nested directories.

Will init-deep overwrite my existing documentation?

By default, the tool is additive and preserves existing content. You can use the --fresh flag if you specifically want to delete existing structures and start over.

Why should I keep CLAUDE.md files under 150 lines?

Claude has an optimal instruction capacity of around 150-200 lines. Exceeding this causes diminishing returns and 'noise' that can lead to the AI ignoring critical project rules.

Nested Context Initializer

Name: Nested Context Initializer
Author: TechDufus

byTechDufus

•

생산성 및 워크플로

Optimizes Claude's contextual awareness by refactoring documentation into a nested, progressive disclosure structure.

소개

The Nested Context Initializer (init-deep) automates the transition from bloated, monolithic CLAUDE.md files to a high-performance nested structure. By strategically placing specific guidance in relevant subdirectories, it ensures Claude only processes the instructions relevant to the current task, preventing instruction fatigue and maximizing the AI's reasoning capacity. The skill uses a proprietary scoring system to identify high-impact directories, ensuring documentation is placed exactly where it provides the most value without exceeding Claude's instruction budget.

주요 기능

Automated directory scoring for optimal documentation placement
Smart deduplication to ensure child files inherit rather than repeat parent rules
Progressive disclosure enforcement to minimize context window bloat
Migration engine to extract content from monolithic files into nested structures
46 GitHub stars
Parallel exploration agents for rapid repository-wide analysis

사용 사례

Bootstrapping new repositories with domain-specific AI guidance for tests and APIs
Refactoring monorepos to provide localized context for individual packages
Migrating legacy projects with oversized, outdated CLAUDE.md files

소개

주요 기능

Automated directory scoring for optimal documentation placement
Smart deduplication to ensure child files inherit rather than repeat parent rules
Progressive disclosure enforcement to minimize context window bloat
Migration engine to extract content from monolithic files into nested structures
46 GitHub stars
Parallel exploration agents for rapid repository-wide analysis

사용 사례

Bootstrapping new repositories with domain-specific AI guidance for tests and APIs
Refactoring monorepos to provide localized context for individual packages
Migrating legacy projects with oversized, outdated CLAUDE.md files