task_games__dots_and_boxes__completable_box_label

Contract

1. Domain: games
2. Scene id: dots_and_boxes
3. Public task id: task_games__dots_and_boxes__completable_box_label
4. Supported query_id values: single
5. Answer schema: option_letter
6. Annotation schema: bbox
7. Program schema: label(filter(candidate_boxes, drawn_side_count(box)=3)); scene=dots_and_boxes; scope=completable_box_label

Program Contract

Program: label(filter(candidate_boxes, drawn_side_count(box)=3)); scene=dots_and_boxes; scope=completable_box_label

Candidate set: the visible game board, pieces, tokens, cards, tiles, marked state, legal-move cues, result panels, and labeled options inside the completable_box_label objective scope. Operands: visible scene state and prompt-bound operands named by filter, candidate_boxes, drawn_side_count, box, dots_and_boxes, completable_box_label. Operation: evaluate label 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 schema; the prompt/annotation contract defines the minimal visual witnesses. Query ids: single.

Reasoning Operations

Families: filtering, counting, state_update

Generation Notes

1. The scene renders exactly six labeled box options A through F.
2. Exactly one labeled option box has exactly three drawn sides and can be completed by drawing one missing side.
3. Query ids are internal replay/sampling keys and do not define public task units.
4. Annotation is the full-cell bounding box of the selected labeled box, projected from the same generated game state used for answer verification.