Fridge Oracle: I Built My Friend a Recipe Helper That Never Leaves the Laptop
This is my entry for the Hacktoberfest Weekend Challenge: Build for a Friend. Demo This is what it looks like start to finish, typed into the real app on my own laptop. I type what is in the kitchen and that I am lactose intolerant, hit submit, Gemma 3 thinks for a bit (it is running on my CPU, no GPU, so this takes under two minutes), and the result comes back correctly flagged because it…
This story is about building a recipe helper called Fridge Oracle, which is designed to pick a recipe from a user's fridge that fits their dietary restrictions. The creator built the app for a friend who has food allergies and lactose intolerance. The app uses open source AI to help with the problem of people cooking with whatever is left in their fridge and not knowing if their restrictions are met.
Fridge Oracle runs Google's Gemma 3 model locally on the user's computer using a tool called Ollama, which ensures that the user's dietary restrictions are never sent to a company's API. This gives the app privacy, being free, offline, and not requiring an API key or rate limit. The app works by typing the kitchen contents and restrictions into the app, and then it submits the information to Gemma 3.
The model generates a recipe, but there is a bug that sometimes outputs an empty string or empty array in fields, so the creator added a filter to treat those phrases as empty.
The creator took the lesson that safety checks should be done by plain code after the model's claim, as the model's own safety claim cannot be the last check if an answer can hurt someone. Additionally, the app is not on a public URL because the main value of the app is to keep the user's allergy information on their computer. The creator has provided clear steps in the README to run the app on their own machine for free using Ollama and one ollama pull command. The full code is available on GitHub.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.