rl-optimization

The rl-optimization skill applies reinforcement learning principles to improve the quality and effectiveness of spec-kit prompts and workflow templates. By providing structured guidance during prompt and template editing, this skill helps users enhance clarity, logical flow, and actionable instructions. It addresses common issues in prompt engineering and template design, such as vague instructions, missing error handling, ambiguous sequences, and incomplete sections, ensuring that workflows and prompts are optimized for consistent performance and usability.

The skill offers core evaluation criteria for both prompts and templates. For prompts, it emphasizes initial clarity, step sequence, action specificity, output guidance, error recovery, and completion signals. For templates, it evaluates section utility, logical order, placeholder clarity, completeness, and conciseness. It also identifies red flags, such as skipped sections, confusing placeholders, redundant content, and poor ordering. Additionally, it provides improvement patterns like explicit over implicit instructions, guided decision-making, and failure recovery to enhance reliability and reduce friction in workflows.

RL-optimization is designed for developers, workflow engineers, and documentation specialists who regularly create or refine spec-kit commands and templates. It is useful when editing command files, updating workflow templates, analyzing user feedback, or reviewing PRs and issues related to prompt clarity and template effectiveness. By ensuring that prompts and templates are both clear and actionable, this skill supports improved workflow outcomes, faster onboarding for new users, and higher overall efficiency in project execution.