Can a prompt replace MyFitnessPal Premium?

Build a calorie and macro logger over open food data, and make database coverage the thing you measure.

Import the Open Food Facts export filtered to your country, plus USDA FoodData Central for whole foods, into SQLite with a full-text index over product and brand names. Normalise everything to per-100 g at import.

Logging is search, quantity, done, with a recents list, a favourites list and saved meals — a saved meal is a named set of foods and quantities logged in one action, and it is how a repetitive breakfast stops costing thirty seconds a day.

Build a coverage report and look at it honestly after a fortnight: the count of foods you had to enter by hand because search found nothing. That number tells you whether the open database works where you shop, and it is the deciding fact for whether this build replaces the subscription or merely resembles it. Make hand entry cheap — name, per-100 g macros, save — and offer to submit new products back to Open Food Facts, since the coverage gap only closes if people fill it.

Daily view: energy and the three macros against targets, with the remainder for the day. Weekly view: the average, which is the number that actually corresponds to how bodies work, and the day-to-day spread.

Barcode scanning in the browser using the device camera and a barcode library is worth the afternoon it costs — it is the single biggest determinant of whether anyone keeps logging.

Do not copy MyFitnessPal's database. Its contents are user-contributed but the compilation is theirs, and scraping it is both a licence problem and a data-quality one, since a large share of its entries are wrong.