aeon-trikaya

aeon-trikaya (GenesisAeon Package 52)

Fractal CREP feedback and Trikaya state classification for numeric-to-symbolic sensemaking. Ported from unified-mandala/GenesisAeonAdvancedAi/.

What this is

A rule-based (not machine-learning) pipeline: numeric input is translated into "symbolic" form (tone frequencies, HSL colors, a glyph), scored with a discrete CREP heuristic, classified into one of three Trikaya states (PRÄSENZ / LEERE / AUFLÖSUNG), optionally refined through a fractal refeedback loop, and persisted with trend/volatility analytics over time. Includes a CLI, a minimal Flask API, and deterministic haiku/poetry generation.

This is a genuine, working precursor to a much larger aspirational "AeonNeuroNetz" vision (neural core, VR, Mandala-UI - see docs/blueprints/) - not that vision itself. See epistemic_status.md for the full, honest breakdown of what's real vs. aspirational, including two corrections to the original extraction prompt (Trikaya is a discrete lookup, not a continuous threshold; one module, crep_eval.py, is a non-functional stub kept for API compatibility).

Install

pip install -e .                    # core
pip install -e ".[plotting]"        # + feedback-graph / mandala visualization
pip install -e ".[flask]"           # + minimal HTTP API

Quick start

from aeon_trikaya import AeonTrikayaSystem

system = AeonTrikayaSystem(depth=3)
result = system.run_cycle(input_values=[0.1, 0.2, 0.3])
print(system.get_trikaya_state())  # e.g. "AUFLÖSUNG"

Or via the CLI (ported as-is):

python -m aeon_trikaya.aeon_cli 0.1 0.2 0.3 --haiku --graph

Diamond Interface

Implements the six-method GenesisAeon Diamond Interface via AeonTrikayaSystem, plus a bonus get_trikaya_state(). See epistemic_status.md for exactly how CREP/UTAC fields are derived from this package's real, discrete scoring (not fabricated to fit the contract).

Ecosystem relationship

Diffed against aeon-ai before extraction (see unified-mandala/MANDALA_MAP.md, 2026-07-26): genuine but modest overlap (~20% of modules have more capable equivalents there); the rest - Trikaya classification, fractal feedback, persistent trend/volatility memory, a plugin loader, visualization, a Flask API - has no aeon-ai counterpart.