skill-creator

name	skill-creator
description	Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Skill Creator

Guide for creating effective skills that extend Claude's capabilities.

For detailed step-by-step process, see references/creation-process.md.

---

What Skills Provide

1. Specialized workflows - Multi-step procedures for specific domains
2. Tool integrations - Instructions for working with specific file formats or APIs
3. Domain expertise - Company-specific knowledge, schemas, business logic
4. Bundled resources - Scripts, references, and assets for complex tasks

---

Core Principles

Concise is Key

The context window is a public good. Only add context Claude doesn't already have.

Challenge each piece: "Does Claude really need this?" and "Does this justify its token cost?"

Prefer concise examples over verbose explanations.

Degrees of Freedom

Match specificity to task fragility:

Freedom	When to Use	Example
High (text instructions)	Multiple valid approaches	Heuristics, guidelines
Medium (pseudocode)	Preferred pattern exists	Configurable workflows
Low (specific scripts)	Fragile operations	Critical sequences

---

Skill Anatomy

skill-name/
├── SKILL.md (required)
│   ├── YAML frontmatter (name, description)
│   └── Markdown instructions
└── Bundled Resources (optional)
    ├── scripts/     - Executable code (deterministic, reusable)
    ├── references/  - Documentation loaded on demand
    └── assets/      - Files used in output (templates, images)

SKILL.md

• Frontmatter: name and description only. Description is the trigger mechanism.
• Body: Instructions loaded after skill triggers. Keep under 500 lines.

Bundled Resources

Type	Purpose	Load Behavior
scripts/	Deterministic code	Execute without reading
references/	Documentation	Load when needed
assets/	Output resources	Use without loading

Don't include: README.md, CHANGELOG.md, installation guides, etc.

---

Progressive Disclosure

Three-level loading system:

1. Metadata (~100 tokens) - Always in context
2. SKILL.md body (<5k tokens) - When skill triggers
3. References (unlimited) - As needed

Key Patterns

Pattern 1: High-level guide with references

## Quick start
[Core example]

## Advanced features
- **Form filling**: See references/forms.md
- **API reference**: See references/api.md

Pattern 2: Domain-specific organization

bigquery-skill/
├── SKILL.md (overview)
└── references/
    ├── finance.md
    ├── sales.md
    └── product.md

Pattern 3: Framework variants

cloud-deploy/
├── SKILL.md (workflow + selection)
└── references/
    ├── aws.md
    ├── gcp.md
    └── azure.md

Guidelines:

• Keep references one level deep from SKILL.md
• Include TOC for files >100 lines
• Avoid duplication between SKILL.md and references

---

Creation Process

1. Understand - Gather concrete usage examples
2. Plan - Identify reusable scripts, references, assets
3. Initialize - Run scripts/init_skill.py <name> --path <dir>
4. Edit - Implement resources and write SKILL.md
5. Package - Run scripts/package_skill.py <path>
6. Iterate - Improve based on real usage

For detailed instructions on each step, see references/creation-process.md.

---

Quick Checklist

• Description includes triggers and "when to use"
• SKILL.md under 500 lines
• Large content moved to references/
• Scripts tested and working
• No unnecessary documentation files
• References linked from SKILL.md with clear guidance