Where automation evolves into cognition.
This repository is the meta-layer for the Hostinger-based AI Systems Lab.
- Working deployments live here โ https://github.com/mindset-dev/local-ai-packaged
- Kubernetes agent workflows (KIND) โ https://github.com/journeyman33/dot-ai (upstream: https://github.com/vfarcic/dot-ai)
- Agent frameworks โ https://github.com/coleam00/ottomator-agents
- Archon (Claude Code RAG server) โ fork: https://github.com/journeyman33/Archon (upstream: https://github.com/coleam00/Archon)
This repo explains the architecture, shows diagrams, and provides lightweight scripts and templates.
It does not duplicate the running Docker Compose stacks.
graph LR
CF["๐ Cloudflare DNS"] --> Caddy["โ๏ธ Caddy Reverse Proxy"]
subgraph Docker_Compose_Projects
L["๐ง localai"]
M["๐งฉ maui"]
MEM["๐งฌ mem0"]
O3["๐ฅ o3 (OpenMRS)"]
A["๐ฎ arcane"]
end
subgraph KIND_Cluster
CP["๐ฅ๏ธ control-plane"]
W1["๐งฑ worker"]
W2["๐งฑ worker2"]
CP --> W1
CP --> W2
end
Caddy --> L
Caddy --> M
Caddy --> MEM
Caddy --> O3
Caddy --> A
Routing: Endpoints are fronted by Caddy and Cloudflare on bigtorig.com.
Projects as Units: Each stack runs with a Compose project name (e.g., -p localai, -p maui, -p o3) so docker compose down only affects that group.
๐งฐ This Repo Contains
/docs โ Architecture notes (network, projects, upstream inspirations)
/scripts โ Non-invasive helpers for documentation/housekeeping
.env.example โ Common variables (no secrets)
What it does not contain: full Docker Compose or secrets. Those stay in local-ai-packaged.