The model-selection report needs the threshold with the best utility under explicit error costs. Evaluate the supported thresholds and apply the defined tie rule when several choices have the same score.
Build the implementation in Python under /app, with /app/run.sh as its entry point. Read one JSON request from standard input and write one JSON result to standard output. Input has nonempty thresholds in [0,1], samples with score in [0,1] and binary label, and nonnegative integer fp_cost and fn_cost. Predict positive when score >= threshold. For each threshold calculate false positives and false negatives. Select minimum total cost, breaking ties with the highest threshold. Return threshold, cost, fp and fn. Reject invalid ranges or empty thresholds.
Reject inputs that violate an explicit validation rule with exit status 2, no standard output, and {"error":"invalid_input"} on standard error. For valid input, exit with status 0. Keep the implementation reusable for new inputs; it must work offline and must not depend on the example data being the only data it receives.
Repair or replace the unfinished /app/run.sh and add source files under /app. Malformed JSON must use the same invalid-input response. Do not read /tests or /oracle from the application.