yance521/super_pe_evaluation

A TDD-style prompt evaluation skill — build reusable eval datasets, run prompt versions against locked test cases, score outputs with evidence, and compare results without overwriting history. Local-first, platform-agnostic, append-only.