What problem does it solve? Running the DLBreach Fortran dam/levee breach model requires deep operational knowledge: card-based input formatting, mixed hour/second unit conventions, Wine-based binary execution on Linux, and coupling with CaMa-Flood. This Skill packages that expertise so an AI agent can build, execute, validate, and couple breach simulations without silent scientific errors. ## Core Features & Use Cases - End-to-end 8-stage pipeline: 11 tools covering installation verification, dam geometry and reservoir curve generation, CaMa-Flood inflow conversion, breach configuration, Fortran binary execution via Wine, 13-column output parsing, breach-flow injection back into CaMa-Flood, and hydrograph/geometry visualization. - Diagnostic recovery: 29 symptom-diagnosis-remedy triplets in diagnostics/triplets.yaml cover known failure modes such as empty output from a missing Downstream_Channel_Flow_Out card, hours-vs-seconds unit corruption, and explicit-scheme timestep instability. - Validation grounding: dag.yaml defines breach_discharge as the rank-1 validation output with cited PBIAS pass bands, and preflight_check.py verifies the binary, Wine runtime, Python environment, and test cases before any run. - Use Case: Given a dam's geometry and reservoir storage curve, extract the CaMa-Flood outflw hydrograph at the dam cell, simulate an overtopping breach with cohesive headcut erosion, then inject the resulting breach hydrograph into CaMa-Flood to route the downstream flood wave. ## Quick Start Run python preflight_check.py in this directory, then ask the agent to build and execute a DLBreach overtopping breach simulation for your dam using the staged tools.