Background Repetitive negative thinking (RNT) is a key transdiagnostic mechanism underpinning depression and anxiety. Using “just-in-time adaptive interventions” via smartphones may disrupt RNT in real time, providing targeted and personalized intervention. Objective This pilot randomized controlled trial evaluates the feasibility, acceptability, and preliminary clinical outcomes and mechanisms of Mello—a fully automated, personalized, transdiagnostic, and mechanistic smartphone intervention targeting RNT in young people with depression and anxiety. Methods Participants with heightened depression, anxiety, and RNT were recruited via social media and randomized to receive Mello or a nonactive control over a 6-week intervention period. Assessments were completed via Zoom sessions at baseline and at 3 and 6 weeks after baseline. Results The findings supported feasibility and acceptability, with high rates of recruitment (N=55), uptake (55/64, 86% of eligible participants), and retention (52/55, 95% at 6 weeks). Engagement was high, with 90% (26/29) and 59% (17/29) of the participants in the Mello condition still using the app during the third and sixth weeks, respectively. Greater reductions in depression (Cohen d=0.50), anxiety (Cohen d=0.61), and RNT (Cohen d=0.87) were observed for Mello users versus controls. Mediation analyses suggested that changes in depression and anxiety were accounted for by changes in RNT. Conclusions The results indicate that mechanistic, targeted, and real-time technology-based solutions may provide scalable and effective interventions that advance the treatment of youth mental ill health. Trial Registration Australian New Zealand Clinical Trials Registry ACTRN12621001701819; http://tinyurl.com/4d3jfj9f