Function-level vulnerability detection models are often trained and tested on apparently equivalent datasets that differ in critical aspects such as class imbalance, the presence of duplicate functions, and contamination across splits. Such differences are often ill-documented and hard to verify. This paper introduces VULTRITION, a lightweight tool for analyzing function-level vulnerability datasets and generating compact summaries as “nutrition labels”. These labels report key facts about a dataset, such as its size, vulnerability-type coverage, class imbalance, redundancy, contamination across splits, and structural characteristics of the functions. These facts may help researchers make informed decisions when selecting datasets for vulnerability detection studies. We used VULTRITION to generate 15 nutrition labels for 9 popular function-level vulnerability datasets, revealing differences not readily apparent in the official dataset documentation. VULTRITION is planned to be extended to support other programming language types and vulnerability datasets beyond functions. We publicly release VULTRITION on GITHUB and as a PyPI package to support independent label generation and extensions. A video demo is available at: https://youtu.be/wEaGWKP5Szo.