skill-authoring

To create reusable agent skills, you structure a SKILL.md file with frontmatter containing a name and description, add concise instructions, and place it in the .agents/skills directory to enable future task automation across sessions.

The SKILL.md format structures reusable skills by defining frontmatter with a name and description for discovery, then guiding authors to include concise instructions, examples, and references to longer content or assets.

You save agent instructions for future tasks by authoring a reusable skill with a SKILL.md file, ensuring the frontmatter includes a name and description, and storing it in a predictable directory structure for consistent access.

Yes, you can link to external references or assets when teaching an agent reusable workflows, allowing you to manage longer content separately while maintaining concise primary instructions within the SKILL.md file.

No dependencies are required to start authoring AI skills, as the process relies on creating a SKILL.md file with proper frontmatter and instructions, then adding it directly to the .agents/skills directory.

The best way to structure agent extensions for task automation is to use a predictable directory structure, define clear SKILL.md frontmatter, and document optional scripts, references, and assets to ensure reliable discovery and execution.