Prompt Engineering & LLM Evaluation Toolkit

Stop guessing whether your prompts work. This toolkit gives you a complete system for prompt engineering: design patterns (CoT, few-shot, output constraining), A/B testing scripts that compare prompts head-to-head, LLM-as-judge evaluation with scoring criteria, regression testing to catch prompt breakage, token cost tracking and optimization, version management, and hallucination detection patterns. Includes ready-to-run bash scripts for every workflow. Built for agent operators and developers who need reliable, repeatable LLM outputs in production.