Skip to content

Skills

Overview

Skills are reusable AI capabilities that reduce cognitive friction by giving models persistent knowledge and tools. Rather than re-explaining the same information in every prompt, you create a skill once and the model can reference it whenever needed.

This section teaches you how to design, structure, and implement skills—from basic templates to advanced cross-platform implementations.

What You'll Learn

Fundamentals

Skill Anatomy - Understanding the components and structure of effective skills
Basic Template (Class A) - Creating straightforward, foundational skills
Designing Tools - How to design tools that work within skills

Advanced Concepts

Advanced Skills (Class B/C) - Complex skills for sophisticated use cases
Semantic Tags - Using semantic tagging for skill organization and discovery
Advanced Deep Dive - In-depth exploration of advanced skill techniques
Common Pitfalls - Mistakes to avoid when building skills

Practical Resources

Semantic Tag Reference - Complete reference for semantic tag usage
Complete Skill Template Example - Full working example to learn from
Fill in the Blank Tool Templates - Ready-to-customize tool templates
Cross-Platform Implementation - Implementing skills across different AI platforms and additional resources.

Why Skills Matter

Every time you explain the same concept, provide the same context, or define the same requirements in a prompt, you're creating cognitive friction. The model must process and understand this information again.

Skills solve this by:

  • Reducing repetition - Define once, reference many times
  • Enabling reusability - Use the same skill across multiple tasks
  • Providing consistency - Models apply the same knowledge reliably
  • Freeing context - Save valuable context window space for actual work

Skill Complexity Classes

Skills are classified by measurable complexity, which determines the appropriate template and architecture:

Class A: Simple Skills

  • Scope: Formatting rules, style guides, organizational constraints
  • Size: < 100 lines typical
  • Tools: Reasoning only OR 1-2 read-only tools
  • Verification: User validates output
  • Architecture: Single-file (SKILL.md only)
  • Examples: "Use Oxford commas in documentation" | "Format dates as YYYY-MM-DD"

Class B: Intermediate Skills

  • Scope: Decision logic, conditional workflows
  • Size: 100-500 lines typical
  • Tools: 2-5 tools, may include state-change tools
  • Verification: Tests or verification scripts helpful
  • Architecture: Single-file with optional references/
  • Examples: "Email formatting with contact directory lookup" | "Code review following team standards"

Class C: Advanced Skills

  • Scope: Complex multi-step workflows, verification-critical
  • Size: 500+ lines (requires multi-file architecture)
  • Tools: 5+ tools, complex orchestration, tool composition
  • Verification: Automated verification essential
  • Architecture: Multi-file (SKILL.md + references/ + scripts/)
  • Examples: "SQL query optimization with automated verification" | "Security code review with vulnerability scanning"

The classification determines which template to use: Class A uses Basic Template, while Classes B and C use Advanced Skills architecture.

Getting Started

New to Skills?

Recommended path:
Skill Anatomy → Basic Template (Class A) → Semantic Tags

Understand the structure first, then build basic skills before advancing.

Need Something Specific?

Key Principle

Skills reduce cognitive friction by making knowledge reusable. Well-designed skills help models work more efficiently and reliably by eliminating repetitive context processing.

Ready to begin? Start with Skill Anatomy →