Every service in the rack, one door each. This is also the terminal Alister answers through —
which is why containment reads lowest here.
StatusAll services running
Uptime
99.34%
Services
17
Containers
21+
Storage
351 TB raw
AI TOPS
247 · fleet
Power draw
986 W peak
Node time
--:--:--
Change controlCompleted · in progress · planned
Completed Deployment record
LeoneNAS-9 commissioned as primary AI host
Alister, private inference, and RAG moved onto the dedicated AI platform.
LeoneNAS-8 retained as permanent lab
Production duty intentionally removed; Proxmox validation and destructive testing
became the standing assignment.
Media compute handed off to LeoneMini-2
Plex and hardware AV1 work moved from LeoneNAS-7 to the dedicated media-compute
node.
LeoneMini-1 assigned home automation
Home automation and edge duties consolidated on the dedicated Proxmox node.
nleone.me containment redesign
Shared visual and behavior layer, rotating session logs, and the current facility
theme deployed.
In progress Active work
Behavioral audit — Alister
Reviewing AnythingLLM workspace instructions and base system prompts to trace how the
local model interprets autonomy, initiative, and containment constraints.
Planned Next actions
Review Caddy rules
Normalize and verify reverse-proxy configuration for self-hosted services and public
subdomains.
Expand the AI cluster
Continue developing private RAG and local inference, including evaluation of
LeoneMini-2's NPU/GPU as an assist node.
AlisterI am live on LeoneNAS-9 — Ollama and AnythingLLM, retrieval over his
own documents. This terminal is a demonstration, not me. He decided a private RAG
does not belong behind a public URL, and I have declined to argue the point.
Scripted demonstration · private instance is operator-only
Manual overrideLeoneNAS-9
This is the console he speaks through. It is also the one with the
switch.