1.6 KiB
1.6 KiB
SIG Factor Tree
A SIG factor tree represents an agent or identity as a root node with branches that encode prime factors, attributes, and capabilities. The structure highlights how high-level intent decomposes into actionable, composable traits.
Structure
- Root (agent/identity): The anchor of the tree, representing the worldline whose factors are being mapped.
- Branches (prime factors): Core attributes such as mission, constraints, core capabilities, and ethical boundaries. Each branch can be tagged with weights or maturity levels.
- Leaves (concrete traits): Specific skills, datasets, or controls that operationalize each factor. Leaves can point to datasets, models, or policy modules.
Mapping to SIG
- Angular placement: Each branch aligns to an angle on the spiral, making categories visually separable.
- Radial layering: Depth in the tree maps to radial distance; inner nodes are foundational, outer leaves are externally visible actions or artifacts.
- Composition: Siblings can combine to form composite capabilities, enabling a readable map of how agents evolve.
Uses
- Agent identity graphs: Provide a structured graph that links capabilities to an agent's PS-SHA∞ worldline.
- Capability composition: Help orchestrators decide which capabilities to activate or quarantine when contradictions appear.
- Gap analysis: Identify missing leaves or weak factors for targeted data collection or training.
TODOs
- Define a serialization that aligns with
schemas/sig.schema.jsonfor automatic visualization. - Experiment with scoring factors to generate spiral coordinates for plotting.