Posts

Engineering the AI Gateway Identity, Guardrails, and Production Hardening in USS When you add AI to a product, the real engineering work isn’t model selection. It’s boundary enforcement. USS uses a Cloudflare Worker as a hardened AI gateway. This post breaks down how it works — and why each layer exists. 1. The Worker Is Not a Proxy A naive implementation would look like: Frontend → Worker → OpenAI That’s not what this is. ...

From Tool to Platform Why I Added AI to Universal Schema Studio Universal Schema Studio didn’t start as an AI product. It started as a practical frustration. XSD to OpenAPI conversions Schema inspection Clean documentation workflows Developer-focused structure It was a tool. Focused. Deterministic. Static. Then something changed. 1. Why Add AI at All? There are two types of AI integration: Feature-driven AI (“because it’s trending”) Workflow-driven AI (“because it reduces friction”) ...

Integrating AI into Universal Schema Studio Architecture & Implementation Deep Dive Universal Schema Studio (USS) started as a static developer tool. Hosted on GitHub Pages, fronted by Cloudflare, entirely client-side. Then I added something that fundamentally changes the risk profile: AI-powered OpenAPI 3.0.3 drafting. This post breaks down exactly how I integrated AI without compromising: Security Cost control Identity Abuse resistance Production discipline This is the architecture-first breakdown. 1. The Starting Constraint: USS Is Static USS runs as: ...

A full walkthrough of how I built a self-hosted view counter, analytics backend, dashboard, and badges for Universal Schema Studio — running entirely on Docker, Cloudflared, and my Raspberry Pi.

A detailed guide on how I built a unified web-based schema viewer that converts XSD and XML files into OpenAPI YAML and renders them with ReDoc — complete with dark mode, drag-and-drop uploads, and live previews.

Step-by-step network monitoring setup

Step-by-step guide to setup OMV on your RPI5

Step-by-step guide to setup Adguard Home

Selecting the right media storage

I have been an avid user of the MagicMirror² framework for almost 6 years now. I have developed, and look after a range of modules. This document will detail these modules and their usage. 🧱 Features Based on debian:trixie for Pi 5 compatibility Installs MagicMirror² v2.32+ Allows custom modules + volume persistence Exposed port for local display 🧪 Custom Modules MMM-Growatt for solar stats MMM-Growatt-Stats for combined stats MMM-MovieListings for movies MMM-NOAA3 for weather-forecast from different providers MMM-Reddit to display content from Reddit MMM-GasMonitor to track your gas levels MMM-NewsAPI custom news feed MMM-Rugby track world rugby rankings and tracks match results for specified leagues MMM-SweepClock a clssic Railway Clock MMM-EskomSePush tracks loadshedding schedule in South Africa MMM-WOTD a word of the day module which caters for multiple languages MMM-PhilipsHue displayes information and status of your Hue lights from your Hue Bridge 🪞 MagicMirror² in Docker — My Complete Setup I recently set up MagicMirror² to run fully containerized in Docker with support for: Mounted custom modules, config.js, and custom.css Server-only mode for remote/browser display Clean build → test → push to Docker Hub workflow Here’s how I did it. ...