We’ve seen a lot of data leakage in proteomics studies—and it matters. Even simple mistakes can produce AUCs >0.7 from pure noise, while making models generalize worse. We built tools to help researchers avoid it. Please share so we can start reporting trustworthy models!