ByteBites — Solution

This is the reference solution for the AI110 Module 2 Tinker ByteBites. Unlike the other Tinkers, ByteBites has no starter repo — students build the whole workspace from scratch. This repo is the finished workspace so you can compare your design and code against a complete example. It is one reasonable solution, not the only correct one.

What's here

File	What it is
bytebites_spec.md	The client feature request + the four candidate classes (Part 1b).
draft_from_copilot.md	The messy first AI diagram, kept to show what the custom agent fixes (Part 1c). Intentionally wrong.
bytebites_design.md	The verified UML class diagram — the real design deliverable (Part 1e).
CLAUDE.md	The custom "design agent" that encodes house rules so the AI stays on-spec (Part 1d).
ByteBites_Design_Reference.md	The provided reference file with its Behavioral Instructions filled in.
models.py	The four classes + the filter / sort / total methods (Parts 2–3).
test_bytebites.py	The pytest suite (Part 4).

Run it

pip install -r requirements.txt   # installs pytest
python models.py                  # runs the manual demo scenario
python -m pytest                  # runs the test suite

The design in one breath

Four classes, related by composition (not inheritance):

• Customer has a purchase_history of Orders.
• Order contains FoodItems and can calculate_total().
• Menu contains FoodItems and can filter_by_category() / sort_by_popularity().
• FoodItem is a plain data object: name, price, category, popularity_rating.

The point of the lab

The design comes first. The most valuable artifact is bytebites_design.md, committed before any Python — and the contrast between draft_from_copilot.md (unconstrained AI) and the final design (AI constrained by CLAUDE.md) is the lesson: you are the architect who says "no" to scope creep and inheritance-where-composition-belongs. If your classes match the spec's four responsibilities, use composition, and your tests actually fail when the code is broken, you've hit the objectives.