Public alpha · Tester guide

Your first ingredient check, step by step.

Start with manual entry. Learn what each verdict means, try the reference examples, and share what you notice.

Begin with text you can inspect.

Barcode and photo/OCR paths are demo-only. They do not verify a real barcode or read your package. Manual entry is the meaningful test path today.

  1. Open the app and check your settings.

    On your first visit, choose Get Started. Open the Settings tab and set Strict Mode OFF for the first four examples below. Review Save Scan History before entering text; saving is on by default when no preference exists.

  2. Choose Enter Ingredients Manually.

    From the home screen, select Enter Ingredients Manually, or use the direct manual-entry link.

  3. Enter the full ingredient statement.

    Type or paste into Ingredient List. For a first test, use water, carrots, salt. For a real label, include the whole ingredient statement, not just the “Contains” declaration. Do not remove unfamiliar ingredients to obtain a clean result.

  4. Select Analyze Ingredients.

    Read the verdict and Why this result. Review Flagged Ingredients, Unresolved evidence when shown, and Ingredients Checked. Confirm that the text being checked matches what you entered.

  5. Repeat with another example.

    Use Check Another Label on the result screen to return directly to manual entry. Or use Done to return home.

Manual Entry screen showing the Ingredient List field and Analyze Ingredients action
Manual entry in the alpha. You supply the text; the app checks it against its current local reference lists.

Four states.
Four different meanings.

The result describes the supplied text under the current rules and your preference. It is not independent product verification.

CLEAN

Every supplied ingredient exactly matches the supported clean reference list. This applies only to that text, not the whole product, its manufacturing process, or certification.

NOT CLEAN

An ingredient matches a current Not Clean rule, or Strict Mode applies your avoidance preference to recognized ambiguity. Some rules are provisional avoidance policies. Read each reason; the headline alone is not proof of an animal source.

DISCERN

A recognized ingredient has an unresolved source that may be plant, microbial, or animal. With Strict Mode off and no other blocking evidence, the app asks you to investigate rather than assume.

UNKNOWN

The text is unsupported, incomplete, or otherwise not fully classified. The narrow vocabulary means many real labels land here. UNKNOWN never means clean.

A NOT CLEAN result can still contain unresolved ingredients. Read the whole explanation and evidence, not just the verdict word.

Try these repeatable checks.

Set Strict Mode OFF, then enter each line separately and compare. These are regression examples for the alpha, not recommendations about what to eat.

01 · Expected: CLEAN

water, carrots, salt

Supported clean terms only. The verdict applies to the exact text supplied.

02 · Expected: NOT CLEAN

pork

A direct match under the current Not Clean rules.

03 · Expected: DISCERN

oats, natural flavors, salt

“Natural flavors” has an unresolved source. This expected result requires Strict Mode OFF.

04 · Expected: UNKNOWN

water, xyzzy extract

Synthetic test text, not food. The unsupported phrase must remain unresolved.

Then test the preference: turn Strict Mode ON in Settings and submit oats, natural flavors, salt again. Expect NOT CLEAN with a Strict Mode explanation. This reflects your avoidance preference—not proof of an unclean animal source. The synthetic UNKNOWN example must remain UNKNOWN.

If your result differs, keep the exact text and Strict Mode setting in your feedback. Do not edit the evidence to match the expected verdict.

Know what your settings change.

Use one app tab while testing privacy preferences. Close other SetApart tabs before deleting history; cross-tab coordination is not implemented.

Strict Mode is a preference.

It treats recognized source-ambiguous ingredients as Not Clean by your avoidance choice. It does not establish their source, classify unknown ingredients, or make the app more certain.

Check the setting before comparing results, then run a new analysis. Include the setting when reporting an issue.

The “Show Scripture References” setting is marked coming soon; do not expect it to change explanations yet.

History stays on this device.

In Settings → Save Scan History, turn saving off to stop new history entries. Turning it off does not delete existing history. To remove saved entries, choose Settings → Delete saved history, then confirm. This also clears temporary results in the current tab; new entries can still be saved if saving remains on. Cloud backups are not live.

If a preference fails to load or save, new history is paused in that session. Use the offered Save Off or Save On retry and resolve the error before leaving; the pause may not survive restarting.

Avoid sensitive or personal text. New result links do not include ingredient text. Temporary results expire after refresh or closing the app; reopen saved History or enter the label again. Older alpha links may still contain text in browser history or bookmarks. Deleting app history does not remove those older links or copied notes. Inspect links and screenshots before sharing. Read the in-app privacy policy.

Test the alpha.
Keep its limits in view.

The purpose of this round is to find unclear wording, unsupported ingredients, unexpected results, and broken flows.

What is still a demo or planned

Barcode lookup and photo/OCR scanning use sample data. They are not real product verification or working text recognition. Paid subscriptions and cloud backups are not live.

Ingredient coverage is limited, fuller context-aware rules remain unfinished, and some current rules use provisional avoidance policies.

Not a health or allergy tool

Do not use these verdicts for medical, allergy, or food-safety decisions. Read the physical label and verify unresolved sources with the manufacturer. The app does not provide independent certification or replace your own study and discernment.

Make your feedback easy to reproduce.

Share your notes and screenshots directly with the person who invited you to test. Remove personal information first.

Include these details

Choose a channel you know works.

The app’s feedback screen helps you prepare a note. Fill in the template and choose Copy prepared note, then paste it into a message to your inviter. Nothing is submitted automatically, and receipt is not tracked.

If automatic copy is unavailable, select and copy the Prepared note manually. Your draft stays during in-app navigation, but reload or closing the app clears it; copy your note before leaving.

A guide to keep.
An invitation to share.

Open a resource now or save a copy. All of these are also linked from the app’s tester resource hub.