task_games__snake__path_outcome_option_label

Contract

1. Domain: games
2. Scene id: snake
3. Public task id: task_games__snake__path_outcome_option_label
4. Supported query_id values: single
5. Answer schema: option_letter
6. Annotation schema: bbox_set
7. Program schema: selection.option_value_match(simulate_snake_move_sequence, visible_result_options); scene=snake; scope=path_outcome_option_label
8. Scalar annotation checked: true

Program Contract

Program: selection.option_value_match(simulate_snake_move_sequence, visible_result_options); scene=snake; scope=path_outcome_option_label

Candidate set: the visible game board, pieces, tokens, cards, tiles, marked state, legal-move cues, result panels, and labeled options inside the path_outcome_option_label objective scope. Operands: visible scene state and prompt-bound operands named by simulate_snake_move_sequence, visible_result_options, snake, path_outcome_option_label. Operation: evaluate selection.option_value_match over the candidate set using the visible game state, rules, legal moves, comparisons, counts, simulations, or option-selection constraints encoded in the program expression; generation enforces a unique final answer. Output binding: answer uses the option_letter schema; generation binds a unique final answer. Annotation witnesses: annotation uses the bbox_set schema; the prompt/annotation contract defines the minimal visual witnesses. Query ids: single.

Reasoning Operations

Families: topology, state_update, matching

Generation Notes

1. Simulate the listed moves in order. The answer is the visible option label for the final head cell or GAME OVER.
2. Annotation is the bbox set for visible in-board cells traversed by the head up to the result.
3. The image always shows options A through D, with one GAME OVER option card.
4. Prompt wording comes from src/trace_tasks/resources/prompts/games/snake/games_snake_v1.json.