creating-agent-skills

To write a Claude Code SKILL.md that triggers reliably, enforce Anthropic specification structure by defining clear YAML frontmatter, name conventions, and description triggers. This ensures Claude discovers and activates the skill correctly during workflow routing.

Progressive disclosure in skill authoring is the practice of keeping SKILL.md concise while pushing deep operational details into reference components. This structure prevents context bloat and ensures Claude executes operational instructions efficiently without parsing unnecessary metadata.

To audit an existing Claude Code skill for structural quality, apply specification-aligned rubrics to check YAML frontmatter formatting, naming conventions, and description triggers. This process identifies missing required tags and formatting rules that hinder skill discovery and execution.

A Claude Code skill fails to activate during workflow routing when its YAML frontmatter lacks required tags or its description triggers do not satisfy discovery requirements. Refining the skill metadata and operational instructions ensures Claude can correctly parse and execute the skill.

Yes, you can use Claude Code skill authoring conventions to debug discovery problems by validating SKILL.md metadata against Anthropic structural constraints. Refining frontmatter formatting and operational instructions resolves execution failures caused by misaligned naming and description triggers.