skill-authoring

This skill guides you through creating user-facing agent skills. For the canonical reference, see agentskills.io.

Skill Structure

A skill is a folder containing a SKILL.md file with metadata and instructions:

my-skill/
├── SKILL.md          # Required: instructions + metadata
├── scripts/          # Optional: executable code
├── references/       # Optional: additional documentation
└── assets/           # Optional: templates, resources

SKILL.md Format

Required Frontmatter

---
name: skill-name
description: Brief description of what the skill does
---

• name: Unique identifier (lowercase, hyphens)
• description: One-line summary (loaded at startup for all skills - a maximum of 1000 characters)

Instructions Body

The body contains markdown instructions that tell the agent how to perform the task.

---
name: my-skill
description: Does something useful
---

# Skill Title

Brief overview of what this skill helps accomplish.

## When to Use

Describe scenarios when this skill applies.

## How to Use

Step-by-step instructions or patterns.

## Examples

Concrete examples demonstrating usage.

## Reference

- [Detailed Reference](references/REFERENCE.md) - Link to additional docs

Progressive Disclosure

Structure skills for efficient context usage:

| Layer | Token Budget | When Loaded | |-------|--------------|-------------| | Metadata | ~100 tokens | At startup (all skills) | | Instructions | < 5000 tokens | When skill activated | | References | As needed | On demand |

Guidelines

1. Keep SKILL.md under 500 lines - Move detailed content to references
2. Front-load key information - Put most important patterns first
3. Use tables for quick reference - Easy to scan
4. Link to references - Don't inline everything

Optional Directories

scripts/

Executable code that agents can run:

scripts/
├── validate.sh       # Validation script
├── generate.py       # Code generator
└── setup.js          # Setup helper

Scripts should:

• Be self-contained or document dependencies
• Include helpful error messages
• Handle edge cases gracefully

references/

Additional documentation loaded on demand:

references/
├── PATTERNS.md       # Common patterns
├── API.md            # API reference
└── EXAMPLES.md       # Extended examples

Keep individual reference files focused. Smaller files = less context usage.

assets/

Static resources:

assets/
├── template.xml      # File templates
├── schema.json       # Schemas
└── diagram.png       # Visual aids

File References

Use relative paths from the skill root:

See [the reference guide](references/REFERENCE.md) for details.

Run the setup script:
scripts/setup.sh

Keep references one level deep. Avoid deeply nested chains.

Skill Categories

Developer Skills (.claude/skills/)

Skills for contributors working on this codebase:

• Command development patterns
• Testing approaches
• API client patterns
• Documentation standards

User-Facing Skills (plugins/*/skills/)

Skills for users of the tool:

• CLI command usage
• Platform-specific patterns (B2C Commerce)
• Integration guides

Writing Effective Skills

1. Start with the User's Goal

## Overview

This skill helps you [accomplish X] by [doing Y].

2. Provide Quick Reference Tables

| Command | Description |
|---------|-------------|
| `cmd1`  | Does X      |
| `cmd2`  | Does Y      |

3. Show Concrete Examples

## Examples

### Basic Usage

\`\`\`bash
b2c command --flag value
\`\`\`

### Advanced Usage

\`\`\`bash
b2c command --complex-flag
\`\`\`

4. Explain When NOT to Use

## When NOT to Use

- Scenario A (use skill-x instead)
- Scenario B (manual approach better)

5. Link to Authoritative Sources

Reference official documentation rather than duplicating it:

## Reference

For complete API documentation, see [Official Docs](https://example.com/docs).

Validation Checklist

Before publishing a skill:

• [ ] Frontmatter has name and description
• [ ] SKILL.md under 400 lines
• [ ] Key information appears early
• [ ] Examples are concrete and runnable
• [ ] Reference links are valid
• [ ] No deeply nested reference chains
• [ ] Tested with target agent

Detailed Reference

• Patterns and Examples - Patterns from B2C skills