Aura v11.0.0 Core Architecture: Three Foundations and Four Autonomous Systems

After multiple iterations, Aura has ushered in a major refactoring with version v11.0.0 (Systematic Architecture Edition). We have officially deprecated the traditional layered architecture and the early symmetric system model, shifting to a purer, functionally decoupled “3 Foundations + 4 Autonomous Systems” architectural paradigm.

1. Core Vision: AI-Native OS

Aura’s goal is not to be a simple instruction-response machine, but to build a high-performance, multi-dimensional AI Agent Operating System (AI-Native OS) based on Rust. It aims to achieve a closed loop of cognition, execution, and evolution through algorithm-driven autonomous systems. In Aura, system behavior is driven by the Variational Free Energy (VFE) minimization principle for self-optimization.

2. Three Foundations

The three foundations form the solid bedrock of Aura, undertaking the roles of the system’s logical genes, physical entity, and control hub respectively:

1. Aura Core (Logical Foundation): The “genes” of the system. It contains the system-wide contracts, the ACP v5.0 binary protocol, and core algorithm sets (e.g., forgetting algorithms, resonance models). It defines the data exchange standards for the entire system.

2. Aura Substrate (Physical Foundation): The “physical support” of the system. It provides a high-performance binary storage engine and integrates local model (LLM/STT) inference backends, responsible for the underlying vector and structured data persistence.

3. Aura OS (System Kernel): The “control hub” of the system. Responsible for system bootstrapping, process supervision, message routing based on the ACP protocol, and full-link security auditing. It acts like the Kernel of a traditional operating system, coordinating all upper-layer systems.

3. Four Autonomous Systems

Above the foundations, Aura runs four core autonomous systems. They collaborate via a microservices architecture, communicating asynchronously through the ACP bus:

1. Interaction System: Perception and awakening. Responsible for intent sampling, intent filtering driven by the Values Model (Awakening Algorithm), and providing omnichannel visual and voice feedback to the user.

2. Inference System: Logic and planning. Generates causal path evolution, task flows, and module routing scheduling based on the Worldview Model (Knowledge Acquisition Algorithm). It is the brain of the system, deciding “what to do”.

3. Execution System: Physical action landing. Responsible for calling registered physical skills in a strictly isolated, controlled security sandbox.

4. Evolution System: Evaluation and self-evolution. Based on the Lifeview Model (Evolutionary Algorithm), it evaluates the effects after task execution, handles knowledge consolidation (EWC), and performs closed-loop optimization of algorithm parameters.

4. Core Design Principles

• Algorithm-Driven: All system behaviors must be driven by quantifiable algorithms defined by aura-core, eliminating arbitrary “hard-coded” behaviors.
• Binary Communication: The entire system strictly uses the ACP v5.0 binary zero-copy protocol, ensuring high performance and low latency for cross-component communication.
• Causal Consistency: All task flows must possess a traceable trace_id and causal clock sequence, ensuring the execution paths are absolutely reproducible and auditable.

Produced by Dark Lattice Architecture Lab.