This guide teaches users how to build skills for Claude. A skill is a folder with instructions that teaches Claude to handle specific tasks or workflows. The document is for developers, power users, and teams who want Claude to follow consistent workflows. It covers technical requirements, best practices for skill structure, and patterns for standalone skills and MCP-enhanced workflows. The…
This guide teaches users how to build skills for Claude. A skill is a folder with instructions that teaches Claude to handle specific tasks or workflows. The document is for developers, power users, and teams who want Claude to follow consistent workflows. It covers technical requirements, best practices for skill structure, and patterns for standalone skills and MCP-enhanced workflows.
The guide starts with fundamentals explaining what a skill is and its core design principles: progressive disclosure, composability, and portability. It describes the three-level system of YAML frontmatter, SKILL.md body, and linked files. It also explains how skills work with MCP using a kitchen analogy where MCP provides tools and skills provide recipes.
Chapter 2 covers planning and design. It instructs readers to start with use cases and identify 2-3 concrete scenarios. It presents three common skill categories: Document and Asset Creation, Workflow Automation, and MCP Enhancement. Each category includes real examples like frontend-design skill and sentry-code-review skill. The chapter also defines success criteria with quantitative and qualitative metrics.
The technical requirements section specifies the file structure with SKILL.md, scripts, references, and assets folders. It details critical rules for naming conventions, YAML frontmatter fields including name, description, license, compatibility, and metadata. It includes security restrictions and examples of good and bad descriptions.
The guide continues with instructions for writing the main SKILL.md content using a template structure with steps, examples, and troubleshooting sections. It covers testing and iteration, distribution and sharing, patterns and troubleshooting, and resources and references in later chapters.