distill

To generate Claude Code skills, you use a distillation process that extracts knowledge from provided domain files and code examples to output a structured SKILL.md and EVALUATION.md. This enforces tooling best practices for specific trigger definitions and robust evaluation cases.

Skill distillation is an automated generation process that extracts knowledge from existing domains and code patterns to create self-contained skill definitions. It solves recurring developer tasks by producing structured files with triggers, descriptions, and evaluation scenarios.

Yes, you can automate skill generation to address recurring code failures by providing representative source code and target tasks. The distillation process creates new, self-contained skills that specifically document how to handle these failures.

Yes, providing code style guidelines is part of the input gathering process for distilling a new skill definition. Along with domain files and representative source code, these guidelines help the background sub-agent apply distillation laws accurately.

The best way to create evaluation scenarios is through an automated distillation process that generates an EVALUATION.md file alongside the skill definition. It uses provided domain knowledge and target tasks to produce robust evaluation cases.

Yes, the skill generation process strictly enforces the 15 Laws of Claude Code Tooling. These laws dictate specific trigger definitions, concise descriptions, and robust evaluation cases to ensure the resulting SKILL.md is well-defined and self-contained.