skill-creation-best-practice

Best practices for agent skill instructions involve authoring concise, safe, and testable guidelines. This skill provides structured patterns to reduce trial-and-error and improve consistency across skill development.

To validate agent skill effectiveness, you implement evaluation loops. This skill defines step-by-step evaluation and iteration workflows to test and iteratively improve skills for reliable performance.

Structure agent skill documentation by applying progressive disclosure, modular references, and versionable SKILL.md content. This ensures information is revealed efficiently without overwhelming the agent context.

Progressive disclosure in skill development is a documentation pattern that reveals information sequentially. It keeps SKILL.md content modular and concise, preventing context overload while maintaining comprehensive skill references.

Yes, the evaluation-first approach applies to maintaining existing agent skills. It guides continuous refinement through evaluation loops and structured documentation updates for consistent team workflows.