Open Source Software (OSS) projects depend on a small number of core developers whose departure can stall development or leave critical code unmaintained. Prior work has characterized and labeled inactivity, yet maintainers still lack practical tools to anticipate departures before overall project quality degrades. This paper contributes a forecasting pipeline, an interactive tool built on it, and an initial evaluation with OSS practitioners. First, we develop a model pipeline that reconstructs per-developer daily activity profiles from public GitHub traces, using project-health, socio-technical-network, and knowledge-distribution dimensions feeding them to an LSTM that frames inactivity forecasting as horizon-aware classification, predicting whether a core developer will become inactive within the next k days. Evaluated on 100+ repositories across 45 organizations (226 core developers, ~870,000 developer-days), the pipeline flags up to 75% of true 14-day inactivity-onset windows within a realistic maintainer alert budget. Second, we present DISTRACK, an interactive tool built on this pipeline that delivers model outputs shows at-risk files, project health and simulates the effect of a predicted departure on a project’s collaboration network. Third, we evaluated DISTRACK through structured interviews with OSS practitioners. The project initially question aimed to determine whether developer inactivity could be reliably predicted and our modeling pipeline demonstrated that such predictions are technically feasible. However, the interviews revealed that technical accuracy does not always translate into practical usefulness for all users. Experienced practitioners, particularly long-term core maintainers of small projects, reported limited Perceived Usefulness because they have deep contextual knowledge accumulated over many years, they often already sense when a fellow maintainer is becoming less active. The interviews suggested that the tool’s predictive insights value, if any, lie with other audiences, such as newcomers, future successors, and members of larger distributed communities, who lack the long-term project insight.