skill-best-practices

Validate agent skill metadata by applying a builder-like workflow with a validation script and reference checklist. This enforces compliance with the agentskills.io specification, checking naming, description, folder layout, and frontmatter to ensure content is lean and deployable.

Progressive disclosure moves large schemas or complex rules to a references directory, auto-loading them only when needed. This keeps the main SKILL.md content lean and safe while maintaining access to detailed validation procedures.

Structure new agent skills by establishing consistent naming, descriptions, and folder layout. Following the standardized five-step procedure ensures the skill unit adheres to agentskills.io standards, making it scalable, discoverable, and deployable.

The skill validation workflow requires no external dependencies. It applies to new and existing skill directories by enforcing compliance through a built-in validation script and reference checklist to ensure safe, deployable SKILL.md content.

The best way to author agent skills is using a standardized, scalable approach with templates and validation scripts. This ensures consistent metadata, structure, and frontmatter validation so new skills are discoverable, lean, and deployable.

Move large schemas or complex rules to the references folder when they would otherwise bloat the main SKILL.md file. This progressive disclosure approach auto-loads them only when needed, keeping the primary content lean and deployable.