ONTOLOGY
Glossary & Ontology Map
Plain-language translations of the governance vocabulary.
arifOS has precise definitions for its governance terms. This glossary maps technical vocabulary to plain meaning.
Core Governance Terms
The human operator who holds veto power over any AI recommendation. 888 is the last step before execution — the human seal of approval. If the human disagrees, the AI does not proceed.
The verdict code for an approved action. Means the recommendation passed all 13 constitutional floors and has been recorded in VAULT999. "Sealed" means it happened and it is traceable.
The action is paused. Something in the recommendation triggered concern — uncertainty, irreversibility, potential harm, or a floor violation. The human judge must review before the action can proceed.
The recommendation failed a hard constraint — a constitutional floor that cannot be waived. The action will not proceed. Requires a complete redesign of the approach, not just revision.
The action is not approved but not rejected. Refine the inputs, clarify the uncertainty, add evidence, or narrow the scope. Re-submit after revision. "Sabar" means patience — the work is not done.
One of 13 constitutional rules that every AI action must satisfy. F01 AMANAH (trustworthiness) through F13 SOVEREIGN (human veto). Floors are not suggestions. They are enforcement constraints.
The anti-manipulation constraint. The AI must not claim consciousness, simulate emotions, manufacture false authority, or attempt to persuade the human to expand its own access or autonomy.
The runtime component that enforces the constitutional floors. Every tool call passes through the kernel. The kernel issues the verdict (SEAL/HOLD/VOID/SABAR) and logs to VAULT999.
The A2A agent interface — aaa.arif-fazil.com — where multiple AI agents coordinate within constitutional bounds. The cockpit gives the human operator visibility and control over distributed agent activity.
The declared purpose of an action. Every tool call in arifOS carries a niat field — the operator's stated intent. The kernel checks whether the action actually serves the declared intent.
A scar is a recorded operational failure — a well that should not have been drilled, a model that was wrong, a warning that was ignored. Scar-weight is the gravity the human operator assigns to that failure when reviewing AI recommendations. The scars make the system honest.
The append-only constitutional ledger. Every SEAL, HOLD, VOID, and SABAR verdict is recorded with a timestamp, actor binding, and chain hash. VAULT999 entries cannot be deleted or retroactively modified.
The component responsible for multi-step task decomposition and dependency graph construction. Breaks complex goals into executable sub-tasks with reversibility annotations.
Trinity Symbols
Trinity context: Human Surface — the sovereign origin of authority.
Nine-Signal context: Governance Machine — PSI layer of the observability stack.
Trinity context: Agentic Intelligence Cockpit — the A2A coordination surface.
Nine-Signal context: Infrastructure layer — DELTA of the observability stack.
Trinity context: Constitution — the living set of constraints.
Nine-Signal context: Intelligence layer — OMEGA of the observability stack.