ai-development claude-code pillar-skills skills type-hub Skills (Agent Skills / Claude Skills) Skills provide AI agents with reusable specialized knowledge, procedures, and scripts . This page organizes related articles by category.
Key Takeaways Skills essence : Reusable "instruction packages" for agentsDesign core : Script deterministic processing in scripts/, use text for AI judgmentTrigger design : description is the most critical element for trigger conditionsCore Design Principles Principle 1: Script Deterministic Processing Processing Type Should Script? Reason Character/line counting Required Same result needed every time Syntax/format checking Required Detect all items without omission File conversion Recommended Output format consistency Text editing/review Not needed AI judgment appropriate Design/structure evaluation Not needed Context understanding required
Principle 2: Prioritize Conciseness Claude is already very smart—only add information Claude doesn't inherently have .
Principle 3: Progressive Disclosure Level 1 : Metadata (~100 tokens, always loaded)Level 2 : SKILL.md body (<5000 tokens, on trigger)Level 3 : Bundled resources (unlimited, on demand)Article Map Introduction & Concepts Implementation & Development Comparison & Selection Practice & Use Cases Learning Path graph LR
A[Quickstart<br/>3-min overview] --> B[Complete Guide<br/>Understanding]
B --> C{Goal}
C -->|Create| D[skill-creator<br/>Guide]
C -->|Install| E[Installation<br/>Guide]
C -->|Use API| F[API Implementation<br/>Guide]
D --> G[Practical<br/>Use Cases]
E --> GGrasp concepts : Claude Skills Complete Guide Create Skills : skill-creator Complete Guide Apply in practice : Practical Use Cases February 17, 2026 23:29:52