prompt-lab

Iterate on LLM prompts with structured evaluation and self-correction. Test prompts against ground truth, compare models, track version history. Self-correction loop sends invalid outputs back to LLM