HRR shadow substrate
Vector-symbolic representations can propose advisory similarity or recall signals, but cannot write canonical memory, belief edges, policy state, autonomy state, or soul integrity.
JARVIS includes ambitious HRR/VSA and spatial mental-world research lanes, but the public status must stay honest: these lanes are derived, advisory, and zero-authority.
Vector-symbolic representations can propose advisory similarity or recall signals, but cannot write canonical memory, belief edges, policy state, autonomy state, or soul integrity.
Derived scene graph over canonical perception. It cannot observe, correct, or mutate perception and returns empty state when canonical spatial data is unavailable.
Validation hard-fails raw vector leaks, authority drift, status-marker drift, and private perception-orchestrator access from P5 code.
The Pi display receives only seven scene scalars, never entities, relations, raw vectors, or authority flags.
Research lanes do not promote themselves into shipped status. They require long-run evidence and explicit operator approval.
No authority shortcut
The public page deliberately labels HRR-derived recommendations as advisory. Raw vectors never enter API responses or LLM articulation, and HRR synthetic exercises must keep side effects at zero.