cronky.net is my personal home on the web. It replaced my old blog (blog.cronky.net, 2007 to 2022) in 2026 and brings together my writing, talks, podcasts, projects, photos and rides in one place. It is built the same way I recommend building sites for clients: lightweight, low carbon, accessible and secure, with AI used responsibly as a helper rather than an author.

How it is built

  • Static site. Generated with Jekyll from Markdown and YAML files kept in Git. There is no database, no server-side code and nothing to patch at runtime.
  • Hosting. Served from Cloudflare Pages.
  • Hand-written code. Plain HTML and Sass, with about 6 KB of vanilla JavaScript used only to enhance the page (dark mode, the photo rotation, the mobile menu and search). The content itself does not depend on JavaScript.
  • Search. Pagefind builds a static index at deploy time and runs entirely in your browser, so searches are never sent to a server.

Lightweight and lower carbon

  • The home page is around 220 KB on first load, and most of that is a single photo.
  • Photos are resized and converted to WebP, and the browser picks the smallest size that fits your screen. Images further down the page only load when you scroll to them.
  • No web fonts, frameworks, trackers or adverts are downloaded. Talk recordings only load from YouTube if you press play.
  • The hosting provider, Cloudflare, is listed in the Green Web Foundation's directory of green hosts.

Accessible

The site aims to meet WCAG 2.2 level AA, in both light and dark mode. It is checked with automated testing and by hand. See the accessibility statement for details and known issues.

Secure and private

The site sets no cookies and runs no analytics. It is served over HTTPS only, with a strict Content Security Policy that blocks third-party scripts. See the privacy notice.

Responsible use of AI

I built this site with the help of an AI coding assistant (Anthropic's Claude Code). I believe AI should augment people, not replace their judgement, so these are the principles I followed:

  • Human guidance. I set the direction, structure and tone. The content is my own work or comes from sources I can verify, such as my old blog, my articles and the podcast feeds of shows I've appeared on.
  • Augmentation, not automation. The AI did the heavy lifting I would otherwise not have had time for: migrating hundreds of old posts, finding the exact link for every podcast episode, checking links and auditing accessibility.
  • Human review. I review changes before they go live. Where a fact could not be confirmed, it is marked for me to check rather than guessed.
  • Documentation for humans and AIs. The source includes a README for people and an AI.md file of instructions for AI assistants: who the site is about, house style, the content model and what not to do. There is also an llms.txt to help AI systems describe me accurately.

Mistakes can still slip through, and the site is still in beta. If you spot one, please let me know.

Credits and reuse

  • Writing and photos are mine and licensed under CC BY 4.0 unless otherwise noted.
  • Cycling heatmaps come from Strava. Map data © Mapbox and © OpenStreetMap contributors.
  • Podcast artwork and organisation logos belong to their respective owners and are used to identify them.