[Generated — not hand-authored.] The connection graph — the corpus's interconnections made navigable. Three families: the structural typed graph the data itself declares (realization chains + bridge edges); authored Maps to… / Distinct from… relations resolved to records where high-confidence; and latent edges discovered from the data (a primitive recurring across domains — the convergence signal). Deterministic.

Structural edges — the typed graph

The model's own internal graph, declared in the data: realization chains (each a domain ↔ bridge ↔ domain sequence) and the bridge edges that cross substrate gaps. This is the spine — how the whole system is wired across substrates.

Realization chains (6)

Bridge edges (11)

Authored edges

38 edges across 36 records — distinct-from ×31, maps-to ×7.

8 of 38 authored targets resolved to a concrete record; the rest are prose references kept as text (honest — a wrong link is worse than none).

Latent edges — primitives recurring across domains

The convergence signal: a primitive name appearing in more than one domain. Nobody authored these as links; the data implies them. The more domains a primitive spans, the more load-bearing the convergence.

22 primitive names recur across ≥2 domains.