A KG is a datom fact-log: every line is e=<entity> a=<attribute> v=<value> [c=<context>].
An edge is just a fact whose v=@ref. Schema, nodes, attrs, contexts all compile to this.
A pack (authored .schema/.nodes/.attrs/.context files) is compiled by a generator into a
flat kg.index. A DSL query engine reads the index (nodes | where … | follow … | diff …).
The reference pack is SIFT — a 19-day project snapshot (130 nodes, 19 daily contexts).
Vyasa KG — ACL (viewer-mask) Engine + UI Handoff Prompt