What is the benefit of a hierarchical AGENTS.md system?

It allows AI agents to focus on specific package context without getting overwhelmed by the entire codebase's tokens, while still understanding the global structure and dependencies.

Does this skill support non-JavaScript projects?

Absolutely. It includes detection logic for Go, Rust, Python, and Java projects by scanning for manifest files like go.mod, Cargo.toml, pyproject.toml, and pom.xml.

How does the skill decide which directories to initialize?

It uses a scoring heuristic that looks for manifest files (like package.json or go.mod), core directories (like src/ or lib/), and file counts to identify the most important sub-projects.

Can I run this on an existing project without losing my notes?

Yes, you can use the --preserve flag to keep existing AGENTS.md files or use the --dry-run flag to preview proposed changes before they are written to disk.

Monorepo AI Context Initializer

Name: Monorepo AI Context Initializer
Author: v1truv1us

byv1truv1us

•

Learning & Documentation

Systematically sets up hierarchical agent context files across monorepo subdirectories to optimize AI agent navigation and performance.

This skill automates the creation and linking of AGENTS.md files throughout complex monorepo structures, establishing a three-tier context system consisting of philosophy, coordination, and specialization layers. By intelligently detecting package boundaries using scoring heuristics and generating localized guidance—including tech stacks, build commands, and code styles—it ensures that AI agents maintain clear project visibility without exceeding token limits. It is particularly useful for teams scaling multi-package projects or microservices who want to onboard AI agents efficiently and maintain consistent documentation standards across a distributed codebase.

Key Features

01Localized guidance generation for tech stacks, build commands, and coding patterns

02Customizable scanning depth and directory exclusion patterns to fit any project size

03Smart detection of package boundaries using project file scoring heuristics

04Automatic generation of hierarchical metadata including level, parent, and philosophy links

056 GitHub stars

06Dynamic updating of root CLAUDE.md to provide a global agent coordination index

Use Cases

01Onboarding new AI agents to complex codebases with clear dependency documentation

02Setting up structured AI guidance for a new multi-package repository

03Improving agent context and navigation for existing large-scale microservices

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add v1truv1us/ai-eng-system monorepo-initialization

For use in Claude.ai and ChatGPT

Key Features

01Localized guidance generation for tech stacks, build commands, and coding patterns

02Customizable scanning depth and directory exclusion patterns to fit any project size

03Smart detection of package boundaries using project file scoring heuristics

04Automatic generation of hierarchical metadata including level, parent, and philosophy links

056 GitHub stars

06Dynamic updating of root CLAUDE.md to provide a global agent coordination index

Use Cases

01Onboarding new AI agents to complex codebases with clear dependency documentation

02Setting up structured AI guidance for a new multi-package repository

03Improving agent context and navigation for existing large-scale microservices