The design, implementation and demonstration of a novel and generic computational forecast framework for multi-scale prediction of extreme sea levels and associated flooding is presented. Denoted Water Information Forecast Framework (WIFF), it integrates process-based models for waves, tides and surges from regional to local scales, predicting the flooding of coastal areas, and supporting the routine and emergency management of coastal resources. WIFF manages the simulations and the real-time monitoring data, archives the data and makes the information available through a WebGIS that targets users with distinct access privileges. Additionally, the web component of WIFF adapts automatically and transparently to any device. WIFF also provides ways to assess the model accuracy and generates tailored products based on model results and observations. WIFF is demonstrated in the prediction of extreme water levels in the Portuguese coast, simulating processes at different scales: at basin scales, waves are simulated in the North Atlantic and in the Portuguese shelf, and sea levels due to tides and atmospheric forcings are simulated in the Northeast Atlantic; at estuarine scales, high-resolution, fully coupled wave/circulation predictions are performed in the Tagus estuary to account for wave-current interactions. User-oriented georeferenced products are generated, including automatic model/ data comparisons, targeting the needs of civil protection agents and combining for the first time an agile, service-oriented platform with high-resolution, process-rich predictions of the Tagus dynamics.