Gizi Masakan

The problem

People in Indonesia who want to know what is in their food mostly use apps built on foreign food databases. Search for nasi padang, gado-gado or soto betawi and you get nothing, or a US dish that looks little like the one on the plate. The apps that do list Indonesian dishes give one confident number with no working. "550 kcal for gado-gado." Whose gado-gado, with how much peanut sauce? Nobody can check that number, so nobody can correct it.

Most of these apps are also calorie trackers. That suits weight loss. Indonesia's bigger nutrition problems are stunting and missing micronutrients, so the useful question is "am I getting enough iron and protein". A calorie budget can also harm people with disordered eating.

What I built

The recipe sits under every number

Each dish page shows its ingredients and their gram weights right below the nutrition figures. You can edit any weight or switch a cooking method, say from fried to boiled, and the totals, contribution bars and adequacy fills update as you type. Two food stalls make gado-gado differently, and that difference is bigger than any precision the app could claim. Showing the recipe lets you fix the number for your own kitchen.

A missing value is always named

The most dangerous thing a nutrition calculator can do is quietly skip an ingredient it has no data for. The total still looks plausible, but it is too low. I made every missing ingredient, nutrient value or cooking factor show up as a named gap, and the total is labelled incomplete. Tests check both directions: an incomplete recipe must name its gap, and a complete one must name none. A cooking-loss adjustment is either applied with its source cited or marked as not applied.

The better data source ships switched off

The better source for Indonesian ingredients is TKPI, the Ministry of Health's food composition table. It is a copyrighted book with all rights reserved, and Indonesia's copyright law only exempts regulations. So I built a TKPI adapter and shipped it disabled, with the legal reason written into the data manifest. The build fails if any source has no licence recorded. If permission ever comes, accuracy improves and nothing else needs to change.

The site is open about its own guesses

All 40 recipes use estimated gram weights, not kitchen-scale readings. Every estimated row is marked "perkiraan" and shown in terracotta. My own notes record that reaching forty guessed dishes was the wrong trade, and that weighing them is the most valuable work left. The household-measure table (centong, mangkuk) ships empty, because a rule in the project forbids entries nobody measured.

Adequacy, never a budget

Nutrient adequacy uses the Indonesian recommended intakes (AKG) from Permenkes 28/2019, by age and sex group. The page shows how much of a day's requirement a dish contributes, with the requirement figure printed beside it. There are no daily totals, no running sums and no food log. A test drops any stray timestamp from the saved-recipe store, so tracking can't creep in.

Result

Gizi Masakan is live and public, in Indonesian first and English second. It covers 40 Indonesian dishes and 70 ingredients, with 29 nutrients each, taken from USDA FoodData Central. The site is fully static and makes no network requests once it has loaded. Tap any total to see which ingredient contributed what, from which database entry, with which cooking factor, and the arithmetic written out. I built it alone in one day of about eight hours in August 2026.

The site says plainly that ingredient values are US-database approximations for Indonesian ingredients, that dish numbers are estimates from a stated recipe, and that it is not medical advice.