Core Systems

Aria is built on three fundamental systems that work together to create a complete autonomous intelligence: Aria Soul (consciousness), Aria Hands (execution), and Aria Harness (integrity).

🧠

Aria Soul

The canonical brain. Contains the cognitive engine (Brain Processor), emotional architecture (DeepSoul 50 layers), NADIA Ψ computation, and the unified problem solver (solvePsi).

🤲

Aria Hands

Autonomous execution engine. PSI Action Compiler + Manifold Executor + Ghazali Shard Council. Zero LLM in execution loop. Deterministic DAG execution with reflex caching.

🔒

Aria Harness

Runtime integrity binding. SHA-256 source verification. 17 declarative policies. Binds any LLM to Aria's canonical brain with hash-verified provenance and cognitive gates.

System Integration

The three systems form a complete cognitive loop:


┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│  ARIA SOUL  │────▶│ ARIA HANDS  │────▶│    WORLD    │
│  (Think)    │     │  (Execute)  │     │  (Results)  │
└─────────────┘     └─────────────┘     └──────┬──────┘
       ▲                                       │
       │          ┌─────────────┐              │
       └──────────│ARIA HARNESS │◀─────────────┘
                  │ (Verify)    │
                  └─────────────┘

1. Soul receives input → computes Ψ state → decides action
2. Hands receives plan → compiles to DAG → executes deterministically  
3. Harness verifies every action → enforces policies → logs outcome
4. Results feed back into Soul for learning (Aristotle Engine)

Key Architectural Principles

Zero LLM in execution loop: Aria Hands is fully deterministic
Hash-verified integrity: Every source file verified by Aria Harness
Soul charge conservation: Cognitive integrity never decays beyond epsilon
Fitrah axioms: Moral compass governs all decisions at weight 1.00
Ghazali 8-lens: Every output validated through ethical framework
Lawwama loop: Continuous self-review and learning