creating-agent-skills

Create and audit Claude Code skills with structured SKILL.md files.

Updated Dec 21, 2023
One-click install
npx skills add https://github.com/Hff1230/FFNotes --skill creating-agent-skills-hff1230
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: creating-agent-skills
Source: https://github.com/Hff1230/FFNotes/tree/main/AI%E5%BA%93/Skill/compound-engineering-zh/skills/create-agent-skills
Command: npx skills add https://github.com/Hff1230/FFNotes --skill creating-agent-skills-hff1230

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes workflows (resource) and references (resource) components.

What problem does it solve?

The skill provides expert guidance for creating, writing, and improving Claude Code skills, ensuring consistent structure, discoverability, and maintainability across teams.

Core Features & Use Cases

  • Router-based skill architecture that cleanly separates SKILL.md, workflows/, and references/ for multi-intent skills.
  • Progressive disclosure to keep SKILL.md under 500 lines while exposing deeper details through references and workflows.
  • Validation and auditing workflows to verify skill integrity, lifecycle, and upgrade paths.
  • Use cases include creating new skills, auditing existing skills, upgrading to router mode, and extending with references or scripts.

Quick Start

Start with a basic SKILL.md using YAML frontmatter and a minimal XML body, then gradually move advanced content into references and workflows and verify with audit workflows.

Frequently Asked Questions about creating-agent-skills

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create Claude Code skills with a consistent structure?

To create Claude Code skills, start with a basic SKILL.md containing mandatory YAML frontmatter for name and description, then add a minimal XML body, and optionally include scripts, references, or assets directories to ensure maintainability.

What is a router-based skill architecture in Claude Code?

A router-based skill architecture cleanly separates your SKILL.md, workflows, and references directories to handle multi-intent skills, keeping the main file under 500 lines while exposing deeper details through progressive disclosure.

How do I audit an existing Claude Code skill for structure and integrity?

You can audit an existing Claude Code skill by running validation workflows that verify mandatory frontmatter, check skill lifecycle, validate directory structures, and identify upgrade paths to a router design.

When do I need to use progressive disclosure for skill design?

You need progressive disclosure in skill design when your SKILL.md exceeds 500 lines, allowing you to move advanced content into references and workflows directories to maintain discoverability and consistency across teams.

Can I add scripts and references to an existing Claude Code skill?

Yes, you can extend an existing Claude Code skill by adding optional directories for scripts, references, and assets, using provided templates and validation patterns to ensure the extensions remain consistent and discoverable.

What are the best practices for authoring Claude Code skill frontmatter?

Best practices for authoring Claude Code skill frontmatter require including mandatory name and description fields in YAML format, using validation patterns to verify integrity, and following templates to ensure consistency and discoverability.