skill-creator

Automate Copilot CLI skill creation by drafting a SKILL.md file, defining frontmatter and trigger phrases, then running an evaluation loop to iteratively test and optimize triggering accuracy. This workflow captures intent and refines descriptions automatically.

Improve Copilot CLI skill triggering accuracy by generating trigger evaluations and iteratively refining the frontmatter descriptions and thresholds. This maximizes usefulness across various prompts and contexts while maintaining auditable validation.

Yes, PyYAML is required to build and evaluate Copilot CLI skills using this workflow. It parses and validates the frontmatter definitions within your SKILL.md files during the automated iteration process.

Progressive disclosure for Copilot CLI skills loads metadata upfront, loads the full SKILL.md instructions on trigger, and loads bundled resources like scripts only as needed during activation. This ensures efficient resource allocation during execution.

Yes, you can bundle optional resources such as scripts, references, and assets when defining Copilot CLI skills. These bundled resources are loaded on-demand during skill activation to support the core instructions.

Limitations when automating Copilot CLI skill iteration include strict adherence to project schemas, avoiding malware generation, and maintaining auditable frontmatter. You must ensure safety standards are met while iterating on trigger thresholds and descriptions.