create-agent-skills

Structure an OpenCode skill directory by creating a SKILL.md file with name and description frontmatter, an inline objective, quick_start, and success criteria. You can add optional workflows, references, and assets to support progressive disclosure and maintainable AI-driven workflows.

Progressive disclosure in skill authoring is the practice of revealing complexity incrementally through frontmatter-driven discovery and explicit reference routing. Use it to maintain scalable, maintainable skills that guide AI without overwhelming the core SKILL.md with excessive inline content.

Validate frontmatter correctness for OpenCode skills by auditing the SKILL.md file against the official specification. Ensure name and description fields are present and accurate, enabling proper discovery and compliance across teams for reliable AI execution.

Use a router pattern for slash commands when managing complex skills requiring explicit reference routing and multiple workflows. Use a single-file pattern for simpler skills where a straightforward SKILL.md with inline objectives and quick_start instructions suffices for the AI workflow.

Yes, this skill authoring guide works for auditing existing skills across teams. It provides best practices for evaluating frontmatter correctness, progressive disclosure, and official specification compliance to ensure reliable AI-driven workflows and maintainable skill directories.

XML-structure in OpenCode skill references requires strict adherence to the official specification to avoid parsing errors. Limitations include potential verbosity and rigid formatting constraints, meaning you should carefully balance explicit reference routing with progressive disclosure for optimal AI comprehension.