A 500-year-old formal taxonomy of cognitive capacities, failure modes, and rectification mechanics.

The specification framework behind Etz Chaim AI is Lurianic Kabbalah — re-framed explicitly for ML readers as what it is from an engineering perspective: a 500-year-old formal taxonomy that enumerates cognitive capacities, failure modes, and rectification mechanics, and maps between them with a level of structural discipline that prior cognitive architectures (SOAR, ACT-R, CLARION, LIDA) never inherited.

The 1 696 primary-source assertions in the corpus are not decorative. They are the generative source of the module set: each of the 37 typed modules, each of the 11 named adversarial probes, each of the 13 rectifiers, is backed by specific citations with E1–E6 epistemic labels and a bidirectional spec↔code audit. When the code is wrong, the spec says so. When the spec is wrong, the code says so.

The names are kept unsanitized. Samael, Gamchicoth, Sathariel, Gamaliel — these are not marketing. They are the names attached to specific failure modes in the source taxonomy, and losing them loses the traceability that makes the architecture falsifiable at all.

Full disclosure on the landing is a feature, not a bug. Targets on record skeptical of mystical framings get the "500-year formal taxonomy" phrasing — in their outreach message, not in a sanitized page.