Validation folds need a reproducible share of each class without losing or duplicating samples. Implement the assignment so the fold order and per-class distribution remain deterministic.
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 positive integer folds and samples with unique string id and string label. Group by label, sort IDs within each group, and distribute each group round-robin beginning at fold zero. Return exactly folds lists, each sorted by ID. This balances each class independently; total fold sizes may differ. Reject duplicate IDs or invalid fold count.
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.