Improving Prompts

This skill systematically refines Claude configuration files, custom commands, and skill instructions by applying official Anthropic best practices for Claude 4.5 models. Instead of making generic or structural changes, it focuses on solving concrete performance failures—such as literalness, verbosity issues, or tool-calling errors—by adding necessary context, explicit instructions, and high-quality examples. It ensures that prompt migrations and optimizations preserve what works while leveraging precise principles like 'motivation-based rules' and 'explicit scope' to improve model reliability and adherence to complex instructions.

주요 기능

사용 사례