Naulus
Home

The method

How this is built, and how to check it

Most fitness and nutrition content asks you to trust the author. Naulus asks the opposite, that you check the source. This page is the contract: where the claims come from, how the evidence is graded, how the citations are verified, and what happens when something here is wrong.

Where the claims come from

An extraction pipeline reads peer-reviewed papers indexed in OpenAlex and pulls out the discrete claims each one makes, the population it studied, the effect it found, and the exact supporting sentence. Each claim is normalized, deduplicated, and tied back to its source paper. Nothing is paraphrased into a claim that the paper does not actually make; every claim carries the quote it came from.

The citation-verification gate

This site exists because of a specific failure. Checking a set of confidently-cited health claims one day, several of the citations either did not exist or did not say what they were said to say. Confident, sourced, and wrong. That is the default state of health content, and it is the thing Naulus is built to not do.

So every source is resolved against a citation registry before it is allowed to back a claim. A real DOI that resolves to a real paper, or it does not ship. The check runs against Crossref, the same registry publishers use to mint DOIs.

98.2% resolved

Of 1,000 sampled citations (the most-cited papers behind these claims), 982 resolve against Crossref REST API. Last checked 2026-06-10. Every claim page links its source by DOI so you can resolve it yourself.

How evidence is graded

A claim being published is not the same as a claim being true. Each finding carries a strength grade based on the study behind it, design, size, and how well it converges with the rest of the literature. Grades are shown on every claim, and the weight of evidence, how much research refutes a claim versus supports it, is shown for the myths.

  • 5StrongMeta-analyses and large, well-controlled trials in convergence.
  • 4GoodSolid randomized or large cohort evidence; minor caveats.
  • 3ModerateMixed or smaller studies; the direction is reasonable, not settled.
  • 2LimitedSparse or low-quality evidence; treat as provisional.
  • 1WeakAnecdote, mechanism-only, or contested.

What this is not

Nothing here is for sale and there are no affiliate links, so there is no incentive to push a supplement or a protocol. It is not medical advice; it is a map of what the research says. And it is deliberately small and growing slowly, a curated set of claims done properly beats a thousand done carelessly.

Corrections

When a claim here is wrong, outdated, or sourced badly, it gets fixed and the change is logged in the lab notes. Found one? Write to hello@naulus.com.