ARIA — Behavioural‑Intelligence Substrate

Version: 1.0.0

ARIA is a deterministic, 7-layer behavioral-intelligence substrate designed for OEM integration. It ingests events, normalizes them, verifies identity, assembles operational context, evaluates policies, synthesizes remediation patches, and gates execution with human approval. Processes metadata‑only inputs and returns stable behavioral decisions with full audit traceability.

This repository contains the complete 7-layer ARIA engine, policy evaluator, patch synthesizer, execution gate, audit system, OEM documentation pack, and packaging configuration.

---

7-Layer Architecture

Layer 1: Ingestion & Normalization

• Raw event ingestion from diverse sources
• Normalization to observation primitives

Layer 2: Convergence

• Time-aligned canonical event construction
• Schema validation and format standardization

Layer 3: Identity Verification

• Cryptographic identity verification (JWT, SPIFFE, OIDC)
• Key management and rotation

Layer 4: Operational Context

• Entity resolution and relationship mapping
• Baseline computation and anomaly detection
• Operational graph assembly

Layer 5: Policy Evaluation

• Behavioral policy evaluation engine
• Decision generation with confidence scores
• Extensible policy framework

Layer 6: Patch Synthesis

• Automated remediation patch generation
• JSON patch construction for policy violations

Layer 7: Execution Gating

• Human approval requirement for execution
• Signature verification
• Audit trail completion

Features

• Deterministic 7-layer behavioral-decision pipeline
• Metadata‑only processing (no raw content ingestion)
• Cryptographic identity verification
• Suppression logic and anomaly detection
• Full audit trail with decision provenance
• Extensible policy framework for domain verticals
• OEM documentation pack
• Python package (pyproject.toml)
• Minimal external dependencies

---

Repository Structure

ARIA/
  ├── VERSION
  ├── CHANGELOG.md
  ├── pyproject.toml
  ├── docs/OEM/
  │   ├── README-OEM.md
  │   ├── INTEGRATION-GUIDE.md
  │   ├── API-SURFACE.md
  │   ├── AUDIT.md
  │   └── LICENSE-OEM.txt
  └── src/aria/
      ├── layer1/        # Ingestion & Normalization
      ├── layer2/        # Convergence
      ├── layer3/        # Identity Verification
      ├── layer4/        # Operational Context
      ├── layer5/        # Policy Evaluation
      ├── layer6/        # Patch Synthesis
      ├── layer7/        # Execution Gating
      ├── regulatory/    # Audit & Reporting
      ├── api/aria_api.py
      └── verticals/     # Domain-specific extensions (Financial, etc.)

---

Dependencies

ARIA requires minimal external dependencies:

• PyJWT>=2.0 — JWT token verification for Layer 3 identity verification
• requests>=2.28 — HTTP client for OIDC provider communication

---

Installation

ARIA requires Python >=3.10.

Install from source:

pip install .

---

Usage

from aria.api import AriaAPI

engine = AriaAPI()
signals = engine.run(metadata)

---

OEM Documentation

See docs/OEM/ for:

• Integration Guide
• API Surface
• Audit System
• Versioning
• Changelog

---

Versioning

ARIA follows semantic versioning. OEMs must pin ARIA to:

>=1.0.0 <2.0.0

---

License

See docs/OEM/LICENSE-OEM.txt.