HLN (Holarchical Logic Network) is an architectural pattern designed to orchestrate autonomous, distributed agents in a secure, predictable, and scalable manner.
Unlike traditional A2A (Agent-to-Agent) networks where behavior is emergent and often chaotic, HLN introduces a structured Holarchy (a hierarchy of autonomous units). It transforms unpredictable AI swarms into reliable, fractal business processes.
The fundamental unit of HLN is the Holon. Every Holon follows a simple pattern:
- The Ghost: The internal logic (an AI model, a script, or a human expert).
- The Shell: A standardized interface that makes the Ghost part of the network using the RXON protocol.
By combining them, we get a predictable building block that can be stacked like nesting dolls (Matryoshka).
The "nervous system" of the HLN. Nodes in the network communicate using the RXON (Reverse Axon) protocol. The superior node (Ghost) never connects to the subordinate node (Shell). Instead, the Shell initiates the connection.
A Reverse Axon is a secure, worker-initiated tunnel:
- Bottom-Up Initiative: The "muscle" (Shell) grows a nerve towards the "brain" (Ghost).
- No Open Ports: Nodes run safely behind NAT, firewalls, or on mobile devices.
- Zero Trust: Built-in support for Identity Chains and cryptographically verified message bubbling across infinite layers.
We build Holarchies, not simple Hierarchies.
- Fractal: A complex cluster of Holons can be wrapped in a single Shell to appear as a simple specialist to a higher-level system.
- Recursive: The pattern (Ghost -> Shell) repeats at every scale, allowing for infinite depth.
HLN treats hardware and expertise as abstract units of "Intent".
- Universal: Any device (GPU, Sensor, Actuator) or human skill is matched by arbitrary
properties. - Smart Matching: Intent is matched to reality using standardized GE (Greater or Equal) logic for numbers.
- The HLN Manifesto: Why we need to tame the chaos of AI agents.
- Specification: The technical details of the protocol.
- Security Model: Connectivity security, mTLS, and Zero Trust.
- Worker Protocol Reference: JSON-level API description for implementers.
- Holarchy & Shell-Stacking: Deep dive into the philosophical architecture.
- Security Extensions: Policy enforcement, recursive protection, and cryptographic audit.
- Reactive HLN: Worker-initiated events, signaling, and bubbling.
- Distributed Tracing: How visibility is maintained across layers.
- Pluggable Transports: Decoupling the protocol from the physical network.
- Physical Holons: Extending the grid into the material world.
- Avtomatika Engine: A production-ready Python implementation of the HLN Orchestrator.
- Worker SDK: Python SDK for building HLN-compliant workers.
- RXON Protocol: Lightweight reverse-connection (RXON) protocol for distributed systems. Supports universal resource telemetry (AI/Robotics/IoT) and strictly typed skill contracts. Works behind NAT/firewalls.
We welcome feedback and proposals! Please see CONTRIBUTING.md for guidelines on how to propose changes to the architecture.
This specification is open-source and licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0).