A long time ago, I worked for the Australian Bureau of Meteorology.

While there, I created many internal scripts/apps using weather data, for fun and internal personal use.

One example was a tiny web app that plotted recent temperature observations, forecasts, and in-office temperature using a little USB temperature probe.

It was very cool and a few members of our team kept it terminally open.

Even today, I still leave the temperature forecast page for my suburb open all day long and refresh it before heading out to the gym, supermarket, whatever.

Anyway, with that in mind, I sat down with gpt4o and built out a tiny html + css + js page that grabs temperature observations and forecasts and presets then in a simple one pager.

You can see it here:

No database needed, it hooks directly into the API.

Here’s a screenshot:

Local Temperature Forecast

Initially, we used a GitHub action to download the json/XML products into the repo and then display + plot the relevant data.

We then found a GitHub repo that summarizes the an unofficial API.

Reading and experimenting with the code lead to a much simpler and cleaner page.

We hosted it on GitHub pages. Here’s the code:

I shared it with a friend, realized he lives in a different suburb and gave him instructions on how to change the hard coded suburb details.

I then came back from my run and ask gpt4o to add in a suburb selector.

I’m very happy with the final result.

Like the forecast page before it, this one-page-app sits open all day long.

Another successful example of chat-driven programming.