HSN Agent Network

An early distributed worker network rebuilt from a private archive of node coordination, shared-state scheduling, and lightweight job dispatch experiments.

Overview

This repo is a sanitized reconstruction of the hsn family found in the E:\ archive. The historical system coordinated multiple machines through shared state stored on an FTP-mounted filesystem. Workers identified themselves, reported availability, watched for a manager role, accepted assigned jobs, and returned to idle after each run.

The public version keeps the architecture and drops the unsafe runtime details:

• no raw FTP credentials
• no mounted network paths
• no desktop editor automation
• no live business payload jobs

What This Repo Represents

• worker identity and registration
• shared-state coordination through files
• heartbeat-based manager monitoring
• lightweight leader election
• job assignment and return-to-idle behavior
• a simple update concept for node fleets

Repo Layout

• docs/architecture.md
• docs/lineage.md
• docs/reconstruction-boundary.md
• src/network/shared_state.py
• src/network/registry.py
• src/network/manager.py
• src/network/worker.py
• src/runtime/launcher.py
• src/jobs/example_job.py
• src/jobs/health_check.py
• src/update/update_client.py
• tests/test_network.py

Running The Demo

python -m src.network.demo

The demo creates a local shared-state workspace, registers two workers, elects a manager, dispatches sample jobs, and prints the resulting snapshot.

Running Tests

python -m unittest discover -s tests -p "test_*.py"

Notes

• this repo is reconstructed from historical private work
• the historical runtime used a shared FTP mount and UI-driven launch flow
• the public version replaces those mechanics with a local filesystem simulator and Python callables