skill-creator

To create a new Codex skill, run scripts/init_skill.py to generate a skeleton, edit the SKILL.md frontmatter with name and description, and run scripts/quick_validate.py to validate. This process ensures proper structure, metadata, and discoverability.

The best way to validate a skill structure is by running scripts/quick_validate.py on the skill directory. This validation tool catches basic structural issues with the SKILL.md frontmatter and resource scaffolding, ensuring consistency across skills before deployment.

You need YAML to define skill metadata and agents through the agents/openai.yaml file. The skill creation process enforces frontmatter hygiene and uses YAML to standardize structure, metadata, and validation across the skill unit.

You can use templates to scaffold optional resources like scripts, references, and assets for a skill. The skill creation tooling provides resource scaffolds and example content to accelerate onboarding and standardize progressive disclosure.

Your skill unit fails validation checks when the SKILL.md frontmatter lacks proper name and description fields, or when optional resources are not organized correctly. The validation tool enforces frontmatter hygiene and structural consistency.

The best way to update an existing skill is by editing the SKILL.md frontmatter and organizing optional resources like scripts and references. Run the validation tool after updates to ensure metadata consistency and discoverability across the extended Codex workflows.