cediackermann/riotsecure
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
No description
31 commits 1 branch 0 tags 103 KiB
Shell 98.7%
Starlark 1.3%
Find a file
cediackermann 6e02572bd7 Replace ping sweep with direct parallel port scan across full subnet 
Ping sweep + wait was the bottleneck: it blocked until all 254 pings
timed out before starting the port check. Now all 254 nc checks fire
simultaneously and we stop polling as soon as the first hit is written,
completing in ~1s instead of ~1 minute.
2026-04-24 11:49:32 +02:00
content Add json content files 2026-04-15 11:20:49 +02:00
modelfiles update modelfile 2026-04-15 13:47:11 +02:00
steps Expose Ollama on all network interfaces via launchd agent 2026-04-24 11:28:08 +02:00
check-status.sh Refactor setup into modular steps; add split-device support 2026-04-24 09:53:33 +02:00
cleanup.sh Refactor setup into modular steps; add split-device support 2026-04-24 09:53:33 +02:00
fetchContent.sh add bash file to update models 2026-04-15 11:36:09 +02:00
README.md Refactor setup into modular steps; add split-device support 2026-04-24 09:53:33 +02:00
riot-sources.txt add source fetch script 2026-04-15 11:14:06 +02:00
setup.sh Bootstrap Homebrew before git clone in all setup scripts 2026-04-24 10:20:31 +02:00
setup_ollama.sh Bootstrap Homebrew before git clone in all setup scripts 2026-04-24 10:20:31 +02:00
setup_onyx.sh Replace ping sweep with direct parallel port scan across full subnet 2026-04-24 11:49:32 +02:00
updateModels.sh add bash file to update models 2026-04-15 11:36:09 +02:00

README.md

RIoT AI Setup Guide

Step-by-step instructions for setting up a Mac Mini from scratch.

Presentation Link

Link

Quick Start

Single device (Ollama + Onyx on the same machine)

bash <(curl -fsSL https://raw.githubusercontent.com/cediackermann/riotsecure/main/setup.sh)

Two devices (Ollama on one machine, Onyx on another)

On the Ollama device — installs Ollama and the RIoT models, no Docker needed:

bash <(curl -fsSL https://raw.githubusercontent.com/cediackermann/riotsecure/main/setup_ollama.sh)

At the end the script prints the exact command to run on the Onyx device.

On the Onyx device — installs Docker and Onyx, auto-detects the Ollama device on the network:

bash <(curl -fsSL https://raw.githubusercontent.com/cediackermann/riotsecure/main/setup_onyx.sh)

The script scans the local network for Ollama. If it can't find it automatically, it will ask for the IP address.

Status & Maintenance

Check everything is working:

cd ~/riotsecure && ./check-status.sh

Update Ollama models:

cd ~/riotsecure && ./updateModels.sh modelfiles

Uninstall:

bash <(curl -fsSL https://raw.githubusercontent.com/cediackermann/riotsecure/main/cleanup.sh)

Manual Steps Reference

The scripts pause and prompt when manual action is required. Here is what to expect at each pause.

Docker Desktop resources

Go to Docker → Settings → Resources and set:

Setting Value
CPU limit MAX
Memory limit 20 GB (or MAX if less than 20 GB available)
Disk usage limit MAX

Hit Apply & Restart.

Onyx installer prompts

  1. Press Enter to acknowledge
  2. Choose 2 for Standard
  3. Press Enter for Edge

Web interface configuration

  1. Open http://localhost:3000 and create an admin account.
  2. Go to Admin Panel → Language Models:
    • Select Ollama as provider and give it a name
    • Set the API base URL to the address shown by the setup script
    • Refresh the model list, select at least the three RIoT models, and click Connect
  3. Go to Admin Panel → Chat Preferences → System Prompt → Modify Prompt, delete the entire prompt, and save.

RAG content upload

Option A — File upload (recommended)

cd ~/riotsecure && ./fetchContent.sh riot-sources.txt

Then in the web interface: Admin Panel → Add Connector → File, upload the files from ~/riotsecure/content, and wait ~30 seconds for indexing.

Option B — URL connector

In the web interface: Admin Panel → Add Connector → Web, add each URL from riot-sources.txt as a separate connector using scrape method Single, and wait ~30 seconds for indexing.