Build a dinner-planning app from pantry and weather data

A non-coder used Codex to make a local meal-planning app that draws on kitchen stock, garden harvests, dietary limits, appliances, and the forecast.

Done with Codex· Intermediate· Mar 2026

What they did
Deirdre Straughan, who does not code, started by writing down the features she wanted. She then had ChatGPT reshape that list into a Codex prompt, and Codex assembled a local Python, FastAPI, and SQLite web app in stages. It ran in short bursts, checked its own work, explained next steps, and waited while she granted terminal and file access. She tried it with real inputs, asked for edits, and later bolted on an Edamam recipe feed plus metric unit conversion.
How it went
After about 20 hours and two complete rebuilds, she had a working MVP that held together better. Early versions became buggy and slow because feature requests kept piling up. Edamam costs $9/month.
Tips
Decide the scope before coding starts, and keep each request narrow. When changes are stacked up too quickly, the agent can miss some of them or make the build less stable.

Try it yourself with Codex

Help me create a small local app that recommends dinner ideas using these inputs: pantry inventory [list items], garden harvest [list items], dietary needs [list], kitchen appliances [list], and current weather [conditions]. Build it in stages with Python, FastAPI, and SQLite, and make it straightforward for me to update the stored lists. After each stage, pause and ask me to approve anything that would permanently change files, install software, delete code, or send data. Once it’s ready, launch it locally, show me three example dinner suggestions using test data, and explain how to start it again later.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on beginningwithi.com ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.