Pages, not a database
Every page you read here is a plain file. A build script writes each one from a small data file, one for each recipe, place, story and lesson, so a page cannot drift away from the facts it was made from. A check fails the build if anyone edits a finished page by hand.
There is no account to make and nothing to log in to. The one outside service the pages call is Google Tag Manager, which runs Google Analytics: it counts visits and a few clicks, using its own cookies, so we can see which recipes people like.
Changed in plain words
The people who run this site are learning to make changes by asking an AI helper in ordinary sentences: add a recipe, fix a step, change a picture. The helper works on its own copy of the site and opens a pull request with a preview.
Nothing goes live until the checks pass and a second reader has looked at the change and approved it. That one step is the whole difference between an idea and a page you can visit.
Checked before it goes out
- A guard reads every page and fails if something is on it that must not be, such as a real company detail or a link out to another site.
- A build check fails if a page does not match the data it was made from.
- A findability check makes sure every sentence on a page can be found and changed in one place.
A scheduled job also looks at the live site every morning, checks that its search and sharing files are in order, and tells us about anything broken.
