The Shape Library

50 verified geometric shapes that form the structural foundation of reasoning.

What shapes are

Shapes are verified geometric patterns that represent structural properties of ideas. When an input arrives, it is mapped against the shape library to find the closest structural match. The quality of this match determines the geometric confidence score.

50 verified shapes

The AGI tier includes 50 verified shapes. Each shape has been structurally validated — it represents a real geometric pattern, not an arbitrary classification. Examples include:

  • triangle_stable — Three-point stability, mutual reinforcement
  • pentagon_recursive — Five-fold recursive structure, self-similar patterns
  • hexagon_layered — Six-layer depth, hierarchical emergence
  • circle_complete — Closure, self-reference, completion

Shape matching

The shape library uses geometric distance metrics to find the closest shape. The result includes:

FieldDescription
closest_shapeThe name of the best-matching shape
geometric_confidenceHow well the input matches this shape (0–1)
gap_scoreDistance between the input and the shape (0–1)
firesWhether the gap detector triggered

ASI tier shapes

The ASI tier adds Tier 2 shapes through the doorway-asi package. These shapes capture deeper structural properties: persistence patterns, emergence structures, and self-referential geometries that the core 50 cannot represent.