How a number gets on this site.

Studies and trial registries. Regulators. Other dosing sites and seller guides. Reddit, forums and public communities. News. Reference works.

Most of what people actually do with peptides is written down by communities and other guides, not in PubMed. Leaving that out would make the site tidy and useless. So we show what people do right next to what studies found, and label which is which.

Everything is published in our own words and our own tables. Facts, doses and protocols are fair to repeat. Someone else's text is not, so we never paste it. We keep a short passage from each source to check the fact against, and never republish it.

Every number carries one of these labels.

• Community-reportedWhat people report on Reddit, forums and other communities. Attributed to where it was said, never presented as proof.
• Competitor guideA dosing guide on another reference site. Often where real-world doses get written down first.
• Vendor guideA guide, label or product page from a seller. Good for vial sizes and mixing, weaker on anything else.
• Clinical studyA published study or trial registry entry. Human or animal; the evidence summary on each page says which.
• RegulatoryA regulator or official body: FDA, EMA, WADA, a prescribing label.
• Media reportNews and press, when the report itself is the fact (a recall, a warning letter).
• Reference workEncyclopedias and drug databases, for names, history and chemistry.

A fact backed by several sources takes the label of the weakest one. A dose cited to a trial and to a forum thread reads “Community-reported”, never “Clinical study”. The order, weakest first: community, media, reference work, competitor guide, vendor guide, clinical study, regulatory.

It means a label can undersell a number but never oversell it. A fact also can't claim an origin that none of its sources has. The schema refuses it.

Sometimes we read a fact somewhere other than where it started. Forum dose reports collected on another guide. A trial result covered in the news. The label follows where the fact started, and the source notes where we read it: “Community-reported (via a competitor guide)”. A news story repeating a trial does not get to count as a trial, and a trial does not lose its label because a reporter wrote it up.

AI agents do the research and write the drafts. They never invent a number. Every number has to point at a source, or the record fails validation and does not publish.

The math is code. Unit conversions, syringe readings, blend splits and supply plans run in a tested TypeScript library with exact fractions, so 0.1 + 0.2 is exactly 0.3 and nothing is rounded until you see it. Tests pin the edges: the smallest syringe marks, mg against mcg slips, vials that run out mid-week.

Counts are code too. When a page says a side effect is commonly reported in a community, a model points at the posts and code counts them, one mention per post. The Peptide Evidence Index is counted the same way.

• Schema validation. Every record is checked before it publishes: every number has an origin and a dated source, the label matches the weakest source, ranges run low to high, units fit the route. One failed check and the record stays unpublished while the last good version keeps serving.
• Source health. Once a week a job opens every source link. A dead or moved source gets flagged against the exact claims it supports.
• Two-model claim checks. Being added now: a second model reads each claim against its stored source passage, and dose changes wait until both agree. Until a record has been through it, that record rests on the checks above.

No doctor, pharmacist or other human expert reviews these pages before they publish, and we never call anything “medically reviewed”. You get sources, labels and dates instead, so you can check any number yourself.

The site is educational information, not medical advice. A calculator shows the arithmetic of what you typed. It cannot tell you whether a dose is right for you.

Records publish daily when something changed. Community research runs weekly, source checks run weekly, and a record's date only moves when its content does. Each peptide page shows when it last changed and keeps a change log.

Found a wrong number, a dead link or a label you think is off? Email corrections@truepeptidesource.com with the page and, if you have one, a better source. Fixes go into the page's change log.