meta

Frontmatter defines your skill's activation and scope. Required fields include name, description, and activation triggers. Use the three-sentence request format, follow the [Action] [Input] to produce [Output] pattern for descriptions, and include toxicity assessment fields (is_toxic, toxic_reason) to generate valid skill JSON.

Skill descriptions must follow a fixed semantic structure: state the action, define inputs, specify outputs. This high-density format ensures reliable activation and predictable behavior. Combine it with frontmatter validation to catch formatting errors before deployment.

Activation failures stem from missing or malformed frontmatter, unclear request scope, or safety metadata gaps. Use the debugging checklist to validate field presence, verify the description follows the [Action] [Input] to [Output] contract, and confirm toxicity fields are properly set.

Request is a concise three-sentence trigger statement; description is a structured explanation using the [Action] [Input] to produce [Output] format that defines what the skill does and what outputs it generates.

Yes. Apply the frontmatter validation pattern and description framework to refine existing skills. The debugging checklist helps identify gaps in activation scope or safety metadata that reduce reliability.

Toxicity fields (is_toxic and toxic_reason) are part of the valid JSON contract for skill definitions. They ensure safety checks are explicit and metadata remains structurally complete for Claude Code processing.