About
When developing with Large Language Models (LLMs), an unbalanced specification—detailed in some areas but vague in others—often leads to unexpected and undesirable outputs, as the LLM fills in the blanks rather than asking for clarification. Spec Score addresses this by meticulously evaluating your software specifications across four critical axes: completeness, clarity, constraints, and specificity. It not only provides individual scores for each axis but also calculates a crucial balance score, indicating how evenly the spec covers all areas. By highlighting the weakest axis and offering concrete tips for improvement, this tool empowers developers to refine their specs iteratively, ensuring a well-rounded and actionable blueprint for LLMs to build from, ultimately leading to more accurate and intended results.
