build-skill

To build OpenCode agent skills with proper YAML frontmatter, use a structured framework providing naming conventions and progressive disclosure guidelines. This ensures correct metadata formatting and consistent agent skill design.

Progressive disclosure in skill development is a design pattern that structures instructions so agents load only necessary context. It minimizes token usage by revealing detailed references only when the specific task requires them.

Yes, you need bash scripting to initialize an OpenCode skill. The initialization process relies on a bash script to generate the minimal viable skill template and validate the directory structure.

To decide which skill type to use for an OpenCode task, consult a skill type decision tree. This evaluates task complexity and guides you toward the appropriate structural pattern for your specific automation requirements.

You can validate OpenCode skill structure before deployment using built-in validation scripts. These scripts check YAML frontmatter formatting and verify that your skill adheres to required naming conventions and progressive disclosure patterns.

OpenCode skills fail to load instructions correctly when YAML frontmatter is misformatted or naming conventions are violated. Following a validated skill template ensures the agent parses metadata and instructions without errors.