Skip to content

Latest commit

 

History

History
91 lines (62 loc) · 3.18 KB

File metadata and controls

91 lines (62 loc) · 3.18 KB

Documentation

Welcome to the Claude Content Skills documentation.

Quick Navigation

Getting Started

Skills Documentation

Philosophy & Design


About Claude Code Skills

Claude Code skills are a mechanism for extending Claude's capabilities in specific domains. Unlike slash commands (which require explicit user invocation), skills are model-invoked - Claude automatically detects when to use them based on conversation context.

Skill Structure

Each skill follows this structure:

skill-name/
├── SKILL.md           # Required: Skill definition with YAML frontmatter
├── reference.md       # Optional: Detailed reference documentation
├── config.example.json # Optional: Configuration template
└── scripts/           # Optional: Supporting scripts

The SKILL.md file is the entry point:

---
name: skill-name
description: Description that helps Claude know when to use this skill.
  Include trigger phrases and use cases.
---

# Skill Documentation

Detailed instructions for Claude when this skill is active...

How Claude Uses Skills

  1. User makes a request
  2. Claude scans available skills' descriptions
  3. If a skill matches the request context, Claude activates it
  4. Claude follows the skill's documentation to respond appropriately
  5. Claude uses the skill's scripts and tools as needed

Development Philosophy

These skills were developed with the Assess-Decide-Do (ADD) Framework, which influenced several design decisions:

Assessment-First Design

Each skill includes preview/dry-run modes because the ADD framework emphasizes thorough assessment before action. You always see what will happen before it happens.

Clear Decision Points

Skills present explicit choices (batch size, thresholds, placement strategies) rather than making assumptions. This respects the user's decision-making authority.

Execution Safety

Once in "Do" mode, skills include progress tracking, backups, and resume capability. Completing tasks matters, and interruptions shouldn't mean starting over.


Contributing to Documentation

Documentation improvements are always welcome. When contributing:

  1. Match the existing tone (direct, practical, no fluff)
  2. Include real examples where possible
  3. Document edge cases you discover
  4. Keep the ADD framework principles in mind

Part of Claude Content Skills