Skill Writing

About

This skill provides a structured framework for authoring high-quality SKILL.md files, ensuring that Claude Code agents receive precise, efficient, and well-tested instructions. By applying a Red-Green-Refactor approach to prompt engineering, it helps developers observe baseline failures, implement minimal guidance to resolve them, and refine documentation against specific quality gates. It enforces best practices such as progressive disclosure, token efficiency, and persuasion principles to optimize agent behavior and reliability while maintaining a concise context window.

Key Features

• 6 GitHub stars
• Rigorous quality gates for frontmatter format and search optimization
• Validation-critical workflow patterns for destructive operations
• Implementation of persuasion principles for discipline-enforcing guidance
• Enforcement of 500-line limits and progressive disclosure patterns
• Test-driven prompt engineering (Red-Green-Refactor) methodology

Use Cases

• Creating new agent documentation or SKILL.md files
• Refining existing agent prompts to resolve specific behavioral failures
• Standardizing documentation across a suite of custom Claude skills