How we keep this accurate
Most references ask you to trust them. We would rather show the machinery. This page describes the automated checks that gate every build and the sourcing rules behind every claim — the same discipline, whether an article is in one language or five.
- 97
- compounds covered
- 5
- languages, in parity
- 784
- primary citations
- 8
- research areas
These figures are computed from the content at build time, so they cannot drift from what the site actually contains.
The checks that run on every build
None of these are aspirations. Each is code; a failure stops the build, so nothing inconsistent can ship.
- 01
Language parity
every buildEvery compound must exist in all supported languages. A missing translation fails the build — a reference where languages silently disagree is dangerous.
- 02
Translation freshness
every buildA translation must mirror its English source, reference-for-reference and section-for-section. If the English article moves ahead, the stale translation fails the build until it is re-synced.
- 03
Source provenance
on demandEvery entry is checked for traceable citations, and a ledger ranks articles weakest-first — so the thinnest sourcing is always visible, not hidden.
- 04
Citation liveness
monthlyEvery reference URL is re-checked. A genuinely dead link (404 / DNS) is flagged; a publisher that merely blocks bots is reported but never mistaken for rot.
- 05
Verify the verifier
by ruleEvery check above is itself tested: it must pass on correct content, catch a deliberately injected defect, and never flag valid data. A miscalibrated tool produces false confidence as easily as a false alarm.
How we source
Coverage is only worth as much as the sourcing behind it. Five rules are binding on every author — human or machine.
No invented data
Every value — a formula, a registration number, a dose, a study — traces to a source. If it is not in the source, we say so. Nothing is fabricated to fill a gap.
Read the original
Our edge is going where English-only references do not: Russian, Chinese, Japanese and other regional literature, retrieved in the original and translated — not summarised second-hand.
Registry-verified
Regulatory facts are checked against the actual national medicines register, not the marketing — which is how we catch that a drug is registered for a different use, or that its approval has quietly lapsed.
Honest grading
Every finding carries what it actually is — human, animal, in-vitro, in-silico, or hypothesis — and is marked reported versus verified. Strong evidence is called strong; thin evidence is called thin.
Symmetric humility
“Not found in the sources we could reach” is never the same as “false.” A great deal of real research is paywalled, regional, or unpublished; absence of evidence is not evidence of absence, and we do not pretend otherwise in either direction.
What this is not
None of this makes the site medical advice, and none of it makes an unproven compound safe. Many compounds described here are not approved for human use in any jurisdiction. Verification means our claims trace to sources and stay internally consistent — it does not mean a treatment works. Always consult a qualified healthcare professional before making decisions about your health.
Found something wrong?
The machinery catches drift and dead links; it cannot catch everything. If you spot an error or have a peer-reviewed source to suggest, email us at info@peptideinfo.wiki — corrections are a feature, not an embarrassment.