skill-writer

To create agent skills that pass validation, follow a canonical workflow enforcing mandatory frontmatter, loading reference guidance, and validating outputs against depth gates and artifact checks.

The best way to structure agent skill authoring workflows is by applying a repeatable process covering resolution, authoring, optimization, and registration to ensure consistent quality.

Yes, you need mandatory frontmatter specifying the skill name and description when authoring agent skills, as this is strictly enforced during the output validation process.

The agent skill creation process loads guidance from references, examples, and patterns to reduce blind spots and ensure the skill aligns with the specification.

Yes, you can update existing agent skills safely by applying the canonical workflow to iteration and evaluation tasks, validating the updated outputs against depth gates and artifact checks.

You should use a structured workflow for agent skill development when you need consistent quality, governance, and reduced blind spots across multiple authoring or registration tasks.