skill-authoring

Structure AI agent skills using a standardized directory layout with SKILL.md, references, and scripts. This enforces consistency, efficiency, and maintainability across repository code generation and workflow automation workflows.

Progressive disclosure in agent skill authoring is a three-level loading system using metadata, instructions, and resources. It optimizes context usage by only loading necessary information for code generation and workflow automation tasks.

Write clear trigger descriptions for SKILL.md files by following authoring discipline templates and rules. This ensures descriptions remain concise, trigger-accurate, and aligned with the repository's standardized format for AI agents.

Yes, you can use scripts for deterministic logic when authoring agent skills. The standardized structure includes a dedicated scripts directory to handle deterministic processing alongside markdown guidelines and domain-specific references.

Review AI agent skills before merging using a validation checklist focusing on structure, content, and intent. This ensures the skill directory adheres to authoring standards, progressive disclosure, and consistent formatting.

Yes, domain-specific references are required to build an effective agent skill. They provide necessary context for AI agents during code generation and documentation tasks, ensuring the skill performs accurately within its target domain.