How do I declare the scope of a new skill?

Every skill should use the YAML frontmatter to explicitly state its scope (e.g., meta-configuration or domain-implementation) and its target (e.g., claude-code-itself or application-feature).

How does the CLAUDE.md hierarchy work?

Precedence follows a specific order: Project-level (most specific) overrides Meta-level, which overrides Global-level (most generic) instructions.

Why do I need to separate meta and domain configurations?

Separation prevents Claude from confusing tool-specific commands (like environment setup) with project-specific functionality (like application deployment), ensuring the AI operates in the correct context.

Is this guide necessary for projects using LangChain or Anthropic SDK?

Yes, it is critical because these SDKs use similar terminology to Claude Code. Without isolation, the AI may fail to distinguish between its own internal tools and the agents you are building.

What is the recommended prefix for Claude Code meta-settings?

The guide recommends using the 'cc-' prefix for all meta-configurations and tools to distinguish them clearly from any application-level code.

Meta-Domain Configuration Isolation

Name: Meta-Domain Configuration Isolation
Author: joaopelegrino

byjoaopelegrino

開発者ツール

Prevents configuration conflicts by isolating Claude Code tool settings from project-specific implementation logic.

概要

This skill provides a comprehensive architectural framework for distinguishing between meta-level configurations (how the AI tool behaves) and domain-level configurations (how the application logic works). By implementing namespace isolation, standardized prefixing conventions, and hierarchical CLAUDE.md structures, it ensures that Claude correctly interprets commands and skills without overlapping contexts. This is particularly vital when developing projects that use agentic SDKs like LangChain or Google ADK, where terminology like 'agent' and 'skill' might otherwise cause functional collisions.

主な機能

Conflict resolution patterns for ambiguous terminology like 'agent' or 'hook'
Explicit scope declaration templates for YAML frontmatter in skill files
Hierarchical CLAUDE.md management across global, meta, and project levels
Context detection checklists to verify directory and precedence before execution
Namespace isolation strategies using mandatory cc- prefixes for meta-configurations
0 GitHub stars

ユースケース

Scaling Claude Code configurations across multiple distinct repositories and environments
Standardizing team-wide naming conventions for custom AI tools and project commands
Developing projects using AI SDKs where 'agent' terminology overlaps with Claude Code defaults

概要

主な機能

Conflict resolution patterns for ambiguous terminology like 'agent' or 'hook'
Explicit scope declaration templates for YAML frontmatter in skill files
Hierarchical CLAUDE.md management across global, meta, and project levels
Context detection checklists to verify directory and precedence before execution
Namespace isolation strategies using mandatory cc- prefixes for meta-configurations
0 GitHub stars

ユースケース

Scaling Claude Code configurations across multiple distinct repositories and environments
Standardizing team-wide naming conventions for custom AI tools and project commands
Developing projects using AI SDKs where 'agent' terminology overlaps with Claude Code defaults