Thanks for catching this. The Cronometer adapter was treating the parsed gram count as a serving multiplier, so a 750g entry got its calories multiplied by 750. The “NaNg” had the same root cause: the portion was stored as the raw string “750.00 g”, which JS coerces to NaN when the diary tries to multiply it for display.

The layout overlap on the duplicate-day dialog is should now be fixed too (added a divider so the buttons have proper visual separation from the radio options).

Both are hopefully now fixed and pushed in rc.14. Grab the latest package, delete the affected day from your diary, and re-import. Items should hopefully now come in with the right values.

Thanks again for the detailed report.

Thanks for implementing it that fast! Unfortunately I wasn’t able to test it.

I found the new section in the readme. So I followed it. I created an admin account (which confused me a bit - my user already had admin permissions. So why is it necessary to create another admin user?). Then I wanted to logout and login with the admin again. But I wasn’t able to figure out how to log out. So I just deleted the cookies and local storage.

After that I was greeted by the login page. So I logged in as admin, entered the User Management - and found an interface to manage users. Ok, sounds logic to find a user management tool in the “user management” menu entry. But the readme said, that there should be oidc settings. Maybe they got lost in a merge conflict or something like that. I was testing on rc12.

Anyways I would prefer an env variable to configure oidc. I like to separate the technical configuration (like credentials, etc) from the user space configuration.

This comment may sound super negative, but I’m actually very grateful you addressed this feature. Thanks!

Thanks for the detailed walkthrough. A few of the rough edges you ran into are fixed in the latest release:

The admin-account confusion makes sense in hindsight. If you were already running NutriTrace single-user, you genuinely didn’t need an admin account, that step exists because OIDC requires user management to be on. The README now spells that out as a prerequisite up front so it doesn’t feel like surprise paperwork.

Logout has a real home now. Profile lives at the very top of Settings as a card showing your name and photo, and Log Out is right there inside it. No more digging through the cookie jar.

The “where are the OIDC settings” question was on me. They were buried inside User Management. They’re now their own top-level section called Authentication, sitting right under User Management, with presets for Authentik, Keycloak, Authelia, Pocket ID, Auth0, Google, and a generic OIDC option for anything else.

And on your env-var preference, that’s in too. You can configure providers entirely from .env or docker-compose.yml. Single-provider shorthand:

OIDC_ISSUER=https://auth.example.com/ OIDC_CLIENT_ID=nutritrace OIDC_CLIENT_SECRET=… OIDC_DISPLAY_NAME=Authentik OIDC_REDIRECT_URIS=https://nutritrace.example.com/api/auth/oidc/callback/1

Multi-provider works with OIDC_PROVIDER_2_, OIDC_PROVIDER_3_, etc. Anything you define this way shows up in the Settings UI with a lock badge and is read-only there, so it’s clear where the source of truth lives. Full doc in .env.example and the README.

Pull the latest and give it another shot when you have a minute. Please confirm the env-var path actually works for your IdP when you have a chance.

Thanks again.

I gave this a shot, but when I press the “preview” button I just get a little popup that says “Invalid CSRF token”.

Hmm… i think i see the issue. The preview / commit upload was missing the CSRF token, so the server was rejecting it before it even read the file. Just pushed a fix. Once you pull it down, hard-refresh the page (Ctrl+Shift+R / Cmd+Shift+R) to grab the new bundle and try again.

OIDC feature has been added to app in latest build (1.0.0-rc9). Please test and let me know how it works for you. I successfully tested with Authentik.

Import feature has been added to app as experimental in latest build (1.0.0-rc9). Please test and let me know how it works for you.

OIDC has been planned, but since now i have had multiple requests, i have started working on it. Will be implemented most likely in next RC release. Keep an eye out!

The JSON-massage route may work. Diary items are self-contained snapshots, IDs in your file are ignored on import, and the only real catch is that re-importing a date overwrites the existing entry. Happy to drop the field-by-field shape if you want to go that route.

That said, native CSV importers for the popular apps (MFP, LoseIt, Cronometer, etc) are now on the near-term roadmap (thanks to your suggestion) as the proper path for this. If you can hold off a bit (and use one of the above), that’ll be much easier.

Thanks for the offers to help with translations. Wanted to share the plan.

I’m wiring up the translation infrastructure now: svelte-i18n with one JSON file per locale in the repo. The workflow once it’s ready is straightforward. There’ll be a single English source file at src/i18n/en.json, contributors copy it to their locale (fr.json, nl.json, de.json, etc.), translate the values, and open a pull request. Keys stay untouched, only values change.

Nothing to do right now. I’ll open a GitHub tracking issue once the source file is stable enough to translate against. A short contributor guide will land with it covering workflow and conventions.

One thing worth flagging early: for nutrition labels specifically, please plan to use the regulatory terms that appear on food packaging in your country rather than the literal English equivalents. So Glucides / Lipides / Protéines for French, Koolhydraten / Vetten / Eiwitten for Dutch, Eiweiß rather than Protein for German, and so on.

More soon.

Quick update on this. Translation infrastructure is live as of v1.0.0-rc.6 (released today). The source file is src/i18n/en.json — about 210 keys covering navigation, page titles, the full auth flow, the onboarding wizard, primary actions in Diary / Foods / Goals / Profile, the AI assistant FAB, and common toasts. Server-side strings (email subjects, push notification bodies) are out of scope for now.

Full workflow is documented in CONTRIBUTING.md → Translations.

About 30% of the client-side surface is extracted in this batch — the screens every user touches every session. Wellness sync messages, deep Settings sub-section labels, and Statistics chart internals are the main gaps. If you hit a screen you use heavily that’s still English while translating, open an issue listing the screen and I’ll pull it forward in the next extraction batch.

Tag me on the PR when you’re ready.

Thanks for the offers to help with translations. Wanted to share the plan.

I’m wiring up the translation infrastructure now: svelte-i18n with one JSON file per locale in the repo. The workflow once it’s ready is straightforward. There’ll be a single English source file at src/i18n/en.json, contributors copy it to their locale (fr.json, nl.json, de.json, etc.), translate the values, and open a pull request. Keys stay untouched, only values change.

Nothing to do right now. I’ll open a GitHub tracking issue once the source file is stable enough to translate against. A short contributor guide will land with it covering workflow and conventions.

One thing worth flagging early: for nutrition labels specifically, please plan to use the regulatory terms that appear on food packaging in your country rather than the literal English equivalents. So Glucides / Lipides / Protéines for French, Koolhydraten / Vetten / Eiwitten for Dutch, Eiweiß rather than Protein for German, and so on.

More soon.

Yes, vibe coded and proud of it. This is a labor of love that I genuinely don’t think I could have completed without Claude Code. The first commit is large because I was working out of a private dev repo and synced everything to the public repo just before the 1.0 RC release. Hope you give it a chance and that it suits your needs. Thanks!

PR tested and merged! Thank you!

Is there a place where we could help with translation ?

I know a few people that would want an app like that but English is not their primary language and won’t bother checking it out at all without some kind of translation.

Great question, and not yet. NutriTrace is English-only right now, and the UI strings are hardcoded throughout the Svelte components. To accept translations I’d first need to wire up an i18n layer (svelte-i18n is the obvious pick) and extract strings to per-locale JSON files. Then translation contributions become straightforward via PRs or something like Weblate/Crowdin. I will add this to my roadmap. Any languages in particular we should prioritize?

Mealie integration allows you to pull in your Meal/Recipe from your self hosted instance of Mealie via api. If you have nutrition facts set there by the recipes total weight, it will pull in that data and then you can set your serving size accordingly so it calculates properly. It also pulls in the recipe pic if set in Mealie.

Barcodes work internationally. i can confirm because i have used in both europe (italy) and africa (south africa). Only difference between nutrition facts i see is that the US uses Sodium and i want to say that at least Europe uses Salt. I have a built in conversion where when one is used the other is automatically calculated.

Yes indeed! On my future to do!

Is the docker image public? I’m getting an “unauthorized” error when attempting to pull it. I’m not getting that error for other ghcr.io images. But I’m not docker expert, so I may be doing something wrong.

Should be fixed now, package was set to private by default on first push. Just flipped it. docker pull ghcr.io/traceapps/nutritrace:latest works without auth now. Thanks for flagging it.

Thanks all, really appreciate the kind words. Feedback is welcome on anything: bug reports, missing features, things that feel rough, or just “this works for my setup.” A few features are flagged Experimental right now and I’d like to harden them enough to drop the badge.

Native Android app is in active development. There’s also a sister project in the works called LiftTrace under the same TraceApps umbrella, same self-hosted Docker setup but for workout tracking (sets, reps, programs, PRs). Not public yet but close.