Behind the build.
Real decisions, useful experiments and lessons from running Luminesca. Welcome to the Journal.
Three hostnames, one Cloudflare Pages project: how we route with a Worker
How a single Pages deployment serves several hostnames, the three traps that cost us the most time, and when this pattern is the wrong choice.
We deleted 51 AI-generated articles: what a content audit actually finds
A generated news archive looked fine by every automated metric and was still fabricated. What we checked, what we found, and the rule we now apply before anything gets published.
Fixing grey-screen Steam screenshots: detection and repair at scale
A grey template band across the bottom of a store screenshot is easy to miss by eye and trivial to detect by code. The formula we used, and what a run over our own image library actually found.
Two hundred travel pages from one data file: the model that made them
How a single JavaScript schema grew into a 200-page destination site across thirty destinations and six page types — and the audit that revealed where the template stopped being enough.
Moving 29 articles from a subdomain to the apex: a migration we did not break
Strong content lived on the wrong domain. How we moved a 56,000-word article library across domains with 301 redirects, sitemap surgery and no duplicates.
Notes from the workbench
These entries document how the network is built and maintained. They cover decisions and incidents at the time of writing, with the context needed to understand them.
About Luminesca →