Architecture Tour
This repository acts as the central hub. Everything that performs actual logic or data capture is a standalone repository (a spoke) linked below. Click any node in the diagram to inspect its role and configuration schema.
Why We Deploy a Hub-and-Spoke Agent Architecture
Context Bleed Prevention
Large monolithic repositories bloat LLM context windows and cause token memory dilution. Keeping tools in small, specialized repositories forces agents to write localized, clean code with zero inter-project side effects.
Lightweight Standalone Remixing
Decoupling workflows into 29 independent spokes allows developers to clone and configure exactly one tool (e.g. Cartoonimator) in minutes, bypassing heavy setup dependencies or project-wide compilations.
Resilient Lifecycle Decoupling
A breaking change or package update in one spoke (like the daily RSS curator) will never block or crash a completely different production spoke (like cartoonimator). Isolation guarantees system resilience.
System Layer Map
connor-stack
The central choose-your-own-adventure documentation portal and setup configuration hub. Houses markdown prompt specifications that visitors can copy and paste into their local environments to clone and configure the remaining spoke repositories automatically.
Configuration Interface
setup-prompts/ ├── 00-orientation.md ├── marketer-quickstart.md └── agent-builder-quickstart.md