Using Software

I stopped looking at the code claude wrote for my physics simulation.

Fiddling with bytes to make satellite image basemaps, or alternatively, the story of how I blindly fumbled around publicly available data, numpy, and claude code. This is a, in a way, part 5 in the ‘(Mis)adventures in AI CLIs’ series. It is not titled as such though because this is more about my own naive python data pipelines mistakes and notes to avoid repeating them in the future than it is about claude code or codex. ...

Airplanes are fun. Maps are fun. Aviation adjacent maps are fun squared? Terrain Flying is the result of a couple months of working ~2–4 hours a day for 2–3 days a week. It’s a barebones prototype, but there is enough there to write about it now. During this time there have been moments where Codex enabled me to do something I never would have bothered to do and there have been moments where I thought this was an incredible waste of time. ...

We continue the prior post’s theme of using CLI AI tools and then writing about it. The theme of Part 2 was directing updates to a legacy javascript library and was based on work done in September, 2025. The theme of this post is on using these tools for brand new development, specifically a financial calculator toy available here. ...

Like previous posts, this was originally going to be about accomplishing something grand. Instead, this is a look into a niche issue that made it hard to build with AI. Specifically, why is it so hard for Codex to run vitest? ...

If Part 1’s theme was attempting to ‘vibe’ through golang text templates with codex then Part 2’s theme is attempting to ‘vibe’ through updates to a legacy javascript library with Gemini CLI. As planned this did not go as planned. I recommend skimming Part 1’s narrative, if not reading, as the following builds on it incrementally. Context, again First, a brief refresh on the origin of this work. I was dissatisfied with an existing hugo theme for my image gallery. I also wanted to present images using modern formats like avif which are unsupported by hugo’s builtin processing capabilities. This had impacts on how the templates for my new theme were structured. It also meant client side, in browser client, processing of image tags (e.g. exif or XMP data) for presentation. ...

Originally, I was going to write about using OpenAI Codex versus Google Gemini CLI to help me code my personal photos website. I could and may still do that, however, here we’ll tell a story about building a Hugo theme with codex, where things went right, and where they did not. We might also take digressions into commentary on the reality of software development and the 1986 Fred Brook’s essay ‘No Silver Bullet…’. ...

We all call it docs or gdoc or Google Docs, but the URL path actually says documents. Regardless, I must be in the minority, but I plead for vector graphics support anyway to whatever team within Google Workspaces is responsible for apps. Raster images Raster images are the jpegs, gifs, pngs, webps, tiffs, avifs, and many other formats of the world. They are of fixed resolution. I, and many people within the software development community before and after me, have created diagrams made of boxes, text, and lines connecting those boxes. We’ve used Visio (does anyone still use Visio?), Lucidchart, diagrams.net, and many other tools to create these images. ...

This is a continuation from a post in April, 2023 on low-code and AI. ...

Should organizations that need to build software care about low code? Should they ever have cared about low code? Will low code be the future of how software is built? Will large language models trained to write code based on text prompts replace all developers? Are these same questions applicable to AI? Do I have answers to these questions? Nope, but I can still editorialize. ...