Editorial Standards
PeptideHound documents what published research reports about peptide research compounds. We are not clinicians and we do not claim clinical authority — the authority on this site is borrowed from the sources we cite, which is why every claim carries one.
PeptideHound Staff · Last editorially reviewed
How a page is built
1. The source pack comes first. Before a word is written we assemble the literature for that compound from PubMed, Europe PMC, ClinicalTrials.gov and the FDA’s enforcement database — full abstracts, registered human trials including terminated ones, and any recall or regulatory action naming the compound.
2. Every identifier is machine-verified. Each DOI is resolved against Crossref and each PMID against PubMed before it can be cited. A source that does not resolve cannot appear on the page. This is an automated gate, not a judgement call.
3. Every source is tiered. We label what kind of evidence each citation actually is — systematic review, human randomised trial, human pilot study, case report, animal model, in vitro, registered trial, regulatory action. The tier comes from the publication record, never from an opinion about the journal.
4. Drafting is AI-assisted, and constrained to the pack. We use AI to draft from abstracts we have already fetched and verified. The model writes from documents in front of it rather than from memory, which is why our citations point to papers that exist. It cannot cite what we did not retrieve.
5. Automated checks run on every page. Before anything publishes we verify that every citation resolves, that every substantive claim sits next to one, that animal findings are labelled as animal findings, and that the page contains no recommended amount, no efficacy claim, and no assertion that a research compound is safe.
6. A person reviews it. Automated checks catch fabricated citations and forbidden language. They cannot confirm that a cited paper actually supports the sentence attached to it. That is a human job and it is where we spend the most time. Safety and protocol pages are reviewed claim by claim against the source.
7. It carries a review date, and we come back. Compound and safety pages are re-reviewed at least every 180 days; vendor records every 90, because business facts change faster than literature.
Why we disclose the AI
Because the honest description is better than the alternative. A lot of health content is written by someone who read a press release. Ours is drafted from abstracts we retrieved, cited to identifiers a machine confirmed resolve, labelled by evidence tier, and checked by a person against the source. We would rather tell you exactly how that works than let you assume something vaguer.
If you find a citation that does not support the claim beside it, that is a failure of our process and we want to hear about it.
What we will not publish
- That any research compound is safe. We report what is documented and what is unstudied.
- A recommended amount, schedule or combination. We report protocols that appear in published studies, with the model and the citation, and never convert an animal amount to a human one.
- A ranking, score or rating of compounds or vendors. A vendor record carries only verifiable business facts and where each was checked — a rule about what may appear in one, not a claim that every check has been run. No finding, status or wording in a record is for sale, and neither is a verification outcome; paid Featured listings are labelled, and what they include is public.
- A claim without a source.
- Anything presented as medical advice.
Sources we use
PubMed and Europe PMC for peer-reviewed literature and abstracts. ClinicalTrials.gov for registered human trials, including terminated ones. The FDA’s enforcement and adverse-event databases for recalls, warning letters and reports. Crossref for DOI verification. WADA and USADA for prohibited-substance status. Company documentation — certificates of analysis — for vendor records only.
We include smaller journals, pilot studies and case reports. For compounds with little human research these are often the only human data that exists, and excluding them would misrepresent the evidence rather than improve it. We label what each one is.
Corrections
Errors are corrected in place, the page’s review date is updated, and the change is logged on our corrections page. We do not quietly edit. Report an error.
Informational and research purposes only. Not medical advice. Many compounds documented on this site are not approved for human use in any jurisdiction. We are not clinicians and this site is not a substitute for professional medical care.
