Generate three family dinner options without uploading pantry data
At a glance
A parent built MasakApa to turn ingredients at home into three locally generated recipe options for his wife. This workflow is listed under Meal Planning, Family, Local AI.
- Agents / tools
- MasakApa, Ollama, Gemma 3 4B, Streamlit
SourceBuilding a tool for these workflows? See weekly rates.
What they did
1) On a computer with Python 3.9+ and Ollama, download nasrul21/MasakApa from GitHub. Start Ollama with `ollama serve` and download the model with `ollama pull gemma3:4b`. 2) In the project folder, create and activate a Python virtual environment, run `pip install -r requirements.txt`, copy `.env.example` to `.env`, and keep the default local Ollama configuration. Start `streamlit run app.py`. 3) In the app, enter available ingredients, choose a cooking mood and set the family serving count. Generate the three suggestions. 4) Compare what each recipe uses and what it still needs, including seasonings. Open the selected recipe and copy or download its cooking instructions.
How it went
The demonstration produces recipe options with cooking steps and separate available, required and optional ingredients.
Worth knowing
Local processing depends on retaining the Ollama configuration. The project also supports hosted providers, which would change where ingredient information is processed.