task_games__battleship__last_ship_cell_label

Contract

1. Domain: games
2. Scene id: battleship
3. Public task id: task_games__battleship__last_ship_cell_label
4. Supported query_id values: single
5. Answer schema: string
6. Annotation schema: point
7. Program schema: label(select(candidate_cells, completes_only_not_sunk_ship)); scene=battleship; scope=last_ship_cell_label

Program Contract

Program: label(select(candidate_cells, completes_only_not_sunk_ship)); scene=battleship; scope=last_ship_cell_label

Candidate set: the visible game board, pieces, tokens, cards, tiles, marked state, legal-move cues, result panels, and labeled options inside the last_ship_cell_label objective scope. Operands: visible scene state and prompt-bound operands named by select, candidate_cells, completes_only_not_sunk_ship, battleship, last_ship_cell_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 string schema; the value is the unique visible candidate-cell label. Annotation witnesses: annotation uses the point schema; the prompt/annotation contract defines the minimal visual witnesses. Query ids: single.

Reasoning Operations

Families: matching

Generation Notes

1. The Battleship scene uses five fleet shapes: Line 5, Line 4, Line 3, Square 2x2, and L 3.
2. This task renders a hidden-ship tracking grid: red hit markers, gray miss markers, fleet-shape panel, and six labeled candidate cells A-F.
3. All non-target ships are fully hit. The target ship has exactly one unhit cell, and exactly one candidate label marks that cell.
4. Annotation is one point at the selected answer cell center.