Math-English Hybrid Notation: A Tool for Tuning LLM Register

via

PromptEngineering

I built an AI chat app for fun, but as I developed, I started getting quite serious about building prompt-testing instruments / CLI tools so that Claude Code could run serious prompt experiments and A/B tests. Most every part of the prompt stack has been scientifically tuned, and from my perspective, it has made the characters very realistic and textured and fun. But I wanted to share one of my key observations, for anyone who's interested in the emerging field of Prompt Engineering: it appears