creating-skills

Create a Claude Agent Skill by generating a validated template with correct YAML frontmatter, naming conventions (gerund form, lowercase, max 64 characters), and a structured SKILL.md file containing sections like When to Use This Skill, Core Functionality, Examples, and Error Handling. Use the built-in validation script to ensure your skill meets Anthropic's specifications before deployment.

Claude Skill names must be in gerund form (e.g., 'creating-skills'), lowercase, and not exceed 64 characters. Avoid reserved words and ensure your skill description stays under 1024 characters. Follow progressive disclosure architecture in your SKILL.md and include proper references like REFERENCE.md and FORMS.md to guide users through end-to-end functionality.

Yes. Use the built-in validation script included in this Skill to check your frontmatter, content structure, and naming constraints against Anthropic's requirements. The script ensures your skill satisfies YAML formatting, description length limits, reserved-word prohibitions, and proper SKILL.md section organization.

A complex Claude Skill should include SKILL.md as the core documentation, plus supporting files like REFERENCE.md for detailed guidance, FORMS.md for structured inputs, and scripts for validation and automation. This multi-file approach enables progressive disclosure and helps users understand scope, usage patterns, and error handling.

Progressive disclosure guides users from high-level purpose to detailed mechanics. Start your SKILL.md with When to Use This Skill, move to Core Functionality and Common Patterns, then include Error Handling and Examples. Separate detailed reference material into REFERENCE.md and structured forms into FORMS.md so users access complexity only when needed.

Simple skill scaffolding generates a single SKILL.md with essential frontmatter and sections. Complex scaffolding adds REFERENCE.md for extended guidance, FORMS.md for structured inputs, and validation scripts. Choose based on your skill's scope: use simple for straightforward tasks, complex for multi-step workflows requiring comprehensive documentation and validation.