darwin-skill

AI skill optimization uses an 8-dimension rubric to evaluate SKILL.md files, scoring structure and effectiveness while utilizing git for iterative improvements to continuously enhance skill documentation quality.

Using version control for iterative skill improvement involves integrating git to track changes during hill-climbing optimization, evaluating AI scoring results, and confirming modifications at critical points before finalizing.

AI scoring evaluates skill structure and effectiveness by applying a detailed 8-dimension rubric to SKILL.md files, measuring specific quality metrics to guide the optimization process and generate actionable improvement scores.

You can include user confirmation steps during the skill evaluation process, as the optimization workflow incorporates human-in-the-loop checkpoints at critical points to ensure quality control before applying changes to SKILL.md files.

Git integration is required to run the 8-dimensional scoring effectively, as it enables hill-climbing optimization by tracking iterative improvements and managing version control throughout the skill evaluation process.