task_puzzles__voxel_cube__cube_visible_projection_count

Public Taxonomy

1. Domain: puzzles
2. Scene id: voxel_cube
3. Source scene: voxel_cube
4. Task id: task_puzzles__voxel_cube__cube_visible_projection_count

Query Contract

1. Supported query_id: single
2. Prompt asks for how many cells are filled in a selected top/front/right orthographic projection, using the rendered front/right direction cue.
3. Internal variation: selected view_direction is trace metadata and a prompt slot.

Program Contract

Program: count(filled_cells(orthographic_projection(stack, view_direction))); scene=voxel_cube; scope=cube_visible_projection_count

Candidate set: the visible voxel stack, unit cubes, projections, changed/reference stacks, and labeled candidate views inside the cube_visible_projection_count objective scope. Operands: visible scene state and prompt-bound operands named by filled_cells, orthographic_projection, stack, view_direction, voxel_cube, cube_visible_projection_count. Operation: evaluate count over the candidate set using the visible states, constraints, transforms, comparisons, counts, paths, or option-selection rules encoded in the program expression; generation enforces a unique final answer. Output binding: answer uses the integer schema; generation binds a unique final answer. Annotation witnesses: annotation uses the bbox_set schema; all filled cells in the target projection grid. Query ids: single.

Reasoning Operations

Families: counting, transformation

Answer And Annotation

1. answer_gt.type = integer
2. annotation_gt.type = bbox_set
3. Annotation schema: unordered bbox_set
4. Annotation target: all filled cells in the target projection grid.
5. scalar_annotation_checked = true.

Prompt Contract

1. Bundle: puzzles_voxel_cube_v1
2. Scene key: voxel_cube
3. Task key: cube_visible_projection_count_query
4. Query key: visible_projection_count