Technical Glossary

State-Locked Protocol

The State-Locked Protocol is our core deterministic mechanism used by DIANA OS to maintain immutable state consistency across logical geometries. It enforces strict computational bounds to prevent structural hallucinations during agentic execution.

Kytin

Kytin is the swarm orchestration layer that governs peer-to-peer logic routing and synchronization across the network. It manages the decentralised autonomous operations between the high-compute nodes and edge devices securely.

Queen Node

The Queen Node is a centralized, high-VRAM compute engine powered by the NVIDIA RTX 6000 Ada Generation. It handles massive relational database compilation into ResinDSL and complex neuro-symbolic tasks within the cluster.

Forager Node

The Forager Node is an air-gapped edge device explicitly powered by the NVIDIA Jetson AGX Thor. It executes mission-critical, localized deterministic inference and utilizes acoustic monitoring for structural verification.

Abstract Thinking Device (ATD)

The Abstract Thinking Device (ATD) is a biological machine architecture that is physically forced to follow rules through kinetic interlocks. It uses a curiosity loop to hunt for logic gaps rather than relying on probabilistic text generation.

Neuro-Symbolic System

The Neuro-Symbolic System is a hybrid AI architecture that combines a neural network "thinking" brain with a symbolic "logical" hardware gate. This ensures probabilistic neural outputs are rigidly filtered through hardcoded logical conditions before execution.

ResinDSL Compiler

Resin DSL is a Deterministic Domain-Specific Language (DSL).

A Deterministic State Graph does not merely describe the operational environment; it defines the mathematical boundaries of legal execution. By compiling high-level agent instructions and massive relational databases into a Deterministic Domain-Specific Language (DSL)—a compiled Abstract Syntax Tree (AST) constraint specification—the schema enforces rigid syntactic rules before a tool is ever invoked.

If a proposed action path does not exist as an explicitly validated node-to-node routing transition within the Deterministic State Graph, the system blocks execution at the compiler level. There are no stochastic detours, token-parsing ambiguities, or unverified tool hallucinations.

Thor-Stinger Architecture

The Thor-Stinger Architecture is a 100Hz hardware interlock system that physically cuts power to edge actuators if the AI acts unpredictably. It guarantees kinetic integrity by serving as a fail-deadly deterministic guard.

Proof of Physical Execution (PoPE)

Proof of Physical Execution is a cryptographic validation protocol that verifies an agentic task was physically computed on designated local NVIDIA hardware geometries (such as the Queen Node or Forager Node) without reliance on external cloud APIs.

AST Sieve Interceptor

The AST Sieve Interceptor is a hardcoded syntactic enforcement layer that intercepts, evaluates, and filters probabilistic LLM outputs against strict Abstract Syntax Tree rules before tool execution.

17 Pillars Schema

The 17 Pillars is a Deterministic State Graph (or Relational Execution Topology). Functionally, it is much more than a standard knowledge graph.

To understand why, you have to look at the massive functional divide between how traditional AI uses a knowledge graph and how DIANA OS utilizes the 17 Pillars Schema.

How Standard AI Uses a Knowledge Graph (Passive & Semantic)

In a conventional AI or RAG (Retrieval-Augmented Generation) setup, a Knowledge Graph is a passive semantic library.

  • Structure: It connects entities using simple Subject-Verb-Object (SVO) triples (e.g., User -> owns -> Laptop, or API -> connects to -> Database).
  • Purpose: When an LLM gets confused, it "reads" the knowledge graph to retrieve background facts and enrich its prompt context.
  • The Weakness: It is purely descriptive and read-only. If an agent hallucinates a destructive command, a standard knowledge graph does nothing to physically stop it—it just sits there holding data.

How DIANA OS Uses the 17 Pillars (Active & Prescriptive)

The 17 Pillars Schema takes the topological structure of a graph (nodes, edges, relationships) and turns it into an active, transactional hardware ledger locked inside the Trusted Execution Environment (TEE).

1. A Transactional State Machine, Not Just a Fact Library

While a standard KG maps concepts, the 17 Pillars maps execution reality. It tracks the exact physical and digital state of your hardware geometries in real time. It knows what the Queen Node (RTX 6000 Ada) is compiling, what the Forager Node (Jetson AGX Thor) is actuating, and where the transaction checkpoints stand.

2. Enforces ACID-Like Rollbacks

The mediator checks the 17 Pillars before committing a transaction. If the AST Sieve Interceptor detects that a physical sensor state contradicts a desktop GUI state, the 17 Pillars schema performs a deterministic rollback (force_state_alignment). A standard knowledge graph cannot execute rollbacks or freeze execution loops.

3. Defines the Permissible Action Space

Instead of just telling an LLM what exists, the 17 Pillars dictates what is mathematically legally allowed to happen. By compiling relational databases into ResinDSL and routing them through the 17 Pillars, you are setting strict syntactic boundaries. If an action path does not exist as a valid node-to-node routing transition within the 17 Pillars, the system blocks the execution at the compiler level.

Geometries

Geometries are the logical data premises and multi-dimensional schemas that organize information within the DIANA ecosystem. They define the boundaries of agent interaction and are secured by the State-Locked Protocol.

Physical AI Execution Runtime (formerly The Apiary)

The Physical AI Execution Runtime is a localized, bare-metal hardware environment and Trusted Execution Environment (TEE) where autonomous agents parse, validate, and execute deterministic commands across physical hardware geometries without cloud latency or external network exposure.

Trusted Execution Environment (TEE)

The Trusted Execution Environment (TEE) is an isolated, cryptographically secure memory enclave within a processor that protects executing code and data from external OS tampering or visibility.

Bare-Metal Edge Runtime

The Bare-Metal Edge Runtime executes directly on the physical NVIDIA Jetson AGX Thor hardware without the latency overhead of virtual machines or cloud containers.