Should I vibe code
Log user-entered nutrition and show transparent nutrient totals without diagnosis
Nutrient data is free. A database where 'chicken breast' means exactly one thing is the entire product.
?
Their verdict, the Gold price and the build-time estimate come from their entry, MIT-licensed. Checked 2026-08-04.
?
Our verdict, the regret score and everything below it. Editorial and unsponsored — nobody can pay to be moved.
The honest answer
why the verdict is what it is
Logging what you ate against a nutrient table is a CRUD app, and the underlying data is free — USDA FoodData Central will hand you the numbers. Cronometer's actual product is curation: one checked entry per food rather than forty user-submitted versions of 'chicken breast' with four different portion sizes and two of them wrong. Build your own and you get a working tracker attached to a food database that degrades a little every time you are in a hurry, plus a slowly growing pile of somebody's health data the moment a second person asks for a login.
What actually breaks
not "if". the specific failures.
- Portion maths. A gram is a gram; 'one medium apple', 'one cup, chopped' and 'one serving' are three separate conversion problems, and getting one wrong produces a day that is silently, plausibly incorrect
- Barcode lookup — the feature everyone asks for first and the one with no free, complete, legally clean source
- Recipe scaling: nutrients that are right per 100g and wrong the moment you divide a tray bake by six
- Targets rendered in the same authoritative typeface as the laboratory-sourced numbers, when they came from a formula off a blog
- Two years of logs, the only genuinely irreplaceable thing in the app, living in a container volume nobody has ever backed up
- The second user, at which point you are holding another identified person's weight, intake and — if you added the field — blood glucose
Is that you?
the verdict is a default, not a law
- It logs your food and nobody else's, on your own machine
- The nutrient numbers come from USDA FoodData Central and you show the source alongside every total
- It reports and never advises — no scores, no targets, no sentences beginning 'you should'
- Anyone else's log is in it
- It computes a target and presents it as a recommendation
- It is aimed at a condition — diabetes, kidney disease, an eating disorder recovery plan — because that is where a wellness app becomes something a regulator has written about
- There is no backup of the log, which is the only part that took years to accumulate
If you build it anyway
the checklist, then the prompt that enforces it
- Build on USDA FoodData Central and store the FDC id with every logged entry, so any total can be traced back to a source you did not invent.
- Store everything in grams internally, with portion descriptions as a display layer. Every unit bug in this category lives at exactly that boundary.
- Report, do not advise. Put the number and the reference intake next to each other and let the reader draw the conclusion.
- If a second person will ever use it, say the sentence out loud first: you are now holding health data about somebody else. Design deletion and export before that day, not after it.
- Back up the log nightly to a different disk, and restore from that backup once, for real.
- Do not scrape a commercial barcode database. The reason there is no free complete one is licensing, which is a contract problem rather than a technical one.
I am building a nutrition and nutrient tracker. Assume it will end up holding health data and apply these before features.
1. Single user by default, running locally. If I ask for accounts or a hosted
version, stop and tell me I am about to hold health data about other
people, and what that changes.
2. Store every quantity in grams or millilitres internally. Portion names are
a presentation layer that converts to grams at the edge. Never store 'one
serving' as a quantity.
3. Every logged item keeps the source database identifier — the FDC id, for
instance — not just the resolved nutrient values. When the source is
revised, my history must still be traceable rather than frozen at whatever
numbers you copied.
4. Build export first: one flat CSV of date, time, food, source id and grams,
runnable before any logging UI exists. Then a nightly backup off the
machine, and a restore I can test.
5. Show provenance in the UI. Every nutrient total displays where its numbers
came from, and anything I typed in myself is visually distinct from
anything laboratory-sourced.
6. Never generate advice. No recommended intakes presented as a
recommendation, no scores, no 'you should', no interpretation of a trend.
Show the number and the published reference value side by side and stop.
7. Refuse to build anything framed around managing a disease. If I ask for
diabetes carb dosing, kidney phosphate limits, or eating disorder
recovery, tell me that reframes the product into a category with its own
regulation and that I should not vibe code it.
8. Do not scrape a commercial barcode or branded-food database. Tell me why
the free complete one does not exist.
9. If accounts ever exist, deletion means the log, the recipes, the weights
and the backups, on request, and it has to be a single tested operation.
10. Out of scope, and say so rather than approximating: wearable sync,
photo-based portion estimation, and any social or sharing feature.That one keeps you out of trouble. For the prompt that actually builds it, canivibecodeit.com has one.
their build prompt ↗Or don’t build it
the boring option, and the way back out
Cronometer's free tier already does full micronutrient tracking against a curated, laboratory-sourced database, which is exactly the part you cannot build. Gold at $10.99 a month adds charts, custom targets and export. If your reason for building was 'I want my own data in a file', check the export button before you check out a repo.
$10.99/mo is cheaper than your weekend.
The log is the asset and it exports as one flat CSV — date, time, food, source id, grams. Everything else re-derives from that. Store the source identifier rather than the resolved nutrient values and the same file still means something in five years, after the underlying database has been revised twice.
Active open-source fitness and nutrition tracker, self-hostable, with the food-database problem already confronted.
Questions
Does HIPAA apply to a nutrition app I wrote?
Almost certainly not. HIPAA binds covered entities — providers, plans, clearinghouses — not a tracker you built for yourself. That is narrower relief than it sounds. The FTC's Health Breach Notification Rule reaches health apps that HIPAA does not cover, and under GDPR anything revealing a person's health is special-category data with its own consent and security requirements. Not-HIPAA is not not-regulated.
Where does a tracker become a medical device?
At the claim, not the code. FDA's general wellness policy leaves low-risk products that talk about general fitness and healthy eating alone. Once a product is presented as helping to diagnose, treat or manage a specific disease, it is in a different category with a different process. A calorie log is fine; the same log rebranded as a diabetes management tool is a different conversation entirely.
Why does the food database matter this much?
Because a nutrition tracker is only as good as its worst entry, and you will never audit it. Curated databases exist because open submission produces forty versions of the same food with different portion sizes, and at the moment you are logging lunch the wrong one is indistinguishable from the right one.
Is the single-user version genuinely fine?
Yes, and it is the version worth building. Your own log, on your own machine, sourced from FoodData Central, reporting rather than advising, is close to a ship-it. Every score on this page is elevated by the two things people add next: other people's accounts, and a target that sounds like advice.
- GDPR Art. 9 — processing of special categories of personal data (EU)
- FDA — General Wellness: Policy for Low Risk Devices (US)
- FTC Health Breach Notification Rule (US) — covers health apps outside HIPAA
Every week, someone ships something they shouldn’t have.
New verdicts, the worst thing that landed in the trap, and the occasional incident report. No other email, ever.
Your scale lies by two kilos and your food log lies low. Yours will not be smart. It will just be confident.
A workout app is a table of sets and a plus-2.5kg rule. The exercise science was never the software.
Four columns and a timer. The only way this hurts is losing four years of them to a cleared cache.
last reviewed 2026-08-04 · verdict is editorial and unsponsored · shared entry data from canivibecodeit under MIT · not legal advice