Life Admin & HomeDone with Codex Mar 3, 2026
Build a household dinner-suggestion app
A non-coder used Codex to build a local app that suggests dinners from pantry stock, garden produce, diets, appliances and weather.
- What They Did
- Deirdre Straughan, a non-coder, wrote a feature list, had ChatGPT turn it into a Codex prompt, and let Codex build a local Python/FastAPI/SQLite web app in small milestones. Codex ran for minutes at a time, tested itself and explained what to try. She approved terminal and file permissions, tested with real data, and requested changes. Later she added an Edamam recipe source and metric conversion.
- How It Went
- After about 20 hours and two full rebuilds, she had a usable, more robust MVP. Ad hoc feature requests had made the first versions buggy and sluggish.
- Worth Knowing
- Agree the scope up front and ask for one change at a time. Ad hoc requests caused bugs, and Codex sometimes ignored all but the first changes. Edamam costs $9/month.
Try it yourself with Codex
Build me a simple local app that suggests dinners based on what's in my pantry ([list items]), my garden produce ([list items]), our diets ([diets]), the appliances we own ([appliances]), and today's weather. Make it easy for me to update the lists. Finish by running it and showing me three sample suggestions, and explain how to start it again.