How do I make my skill token-efficient?

Follow the Progressive Disclosure principle: keep the SKILL.md file lean (under 500 lines) and move detailed references or large schemas into separate files that Claude only loads when necessary.

When should I use low degrees of freedom in a skill?

Use low freedom (specific scripts or strict sequences) when operations are fragile, error-prone, or where consistency is absolutely critical to the outcome.

What is a Claude Code Skill?

A specialized capability that enhances Claude's behavior by providing domain-specific guidance, implementation patterns, and tool integrations for specific tasks.

What is the anatomy of a skill?

Every skill requires a SKILL.md file with YAML frontmatter metadata and markdown instructions. It can optionally include scripts/, references/, and assets/ directories.

What are bundled resources in a skill?

Bundled resources include scripts for deterministic tasks, references for deep domain knowledge, and assets like templates or icons used in Claude's output.

Skill Creator

Name: Skill Creator
Author: januarfonti

byjanuarfonti

开发者工具

Guides the development and optimization of custom Claude Code skills to extend AI capabilities with specialized knowledge and workflows.

关于

The Skill Creator provides a comprehensive framework for building modular, high-efficiency skills that transform Claude into a domain-specific agent. It emphasizes the principle of 'Progressive Disclosure' to manage the context window effectively, offering standardized patterns for organizing documentation, executable scripts, and static assets. By following these best practices, developers can create tools that are token-efficient, deterministic, and highly specialized for complex tasks ranging from specialized API integrations to proprietary business logic.

主要功能

Standardized framework for SKILL.md creation and metadata optimization
Heuristics for setting appropriate degrees of freedom for AI tasks
0 GitHub stars
Progressive Disclosure design patterns to minimize context window bloat
Guidance on bundling reusable scripts, references, and static assets
Step-by-step skill initialization and packaging workflow

使用场景

Optimizing existing skills to be more token-efficient and modular
Building tool-integrated skills with pre-written scripts for data processing
Creating custom onboarding guides for company-specific coding standards

关于

主要功能

Standardized framework for SKILL.md creation and metadata optimization
Heuristics for setting appropriate degrees of freedom for AI tasks
0 GitHub stars
Progressive Disclosure design patterns to minimize context window bloat
Guidance on bundling reusable scripts, references, and static assets
Step-by-step skill initialization and packaging workflow

使用场景

Optimizing existing skills to be more token-efficient and modular
Building tool-integrated skills with pre-written scripts for data processing
Creating custom onboarding guides for company-specific coding standards