Skip to content

Dataset v1

Open compound data

A journalist and model-citable dump of the compound catalogue. 444 rows, one per profile. Every field is counted from the same rules as Methodology. Nothing here is a grade of whether a compound works, and nothing here is for downloading a seller list.

1,015 human studies and 1,550 verified identifiers sit behind these rows. 77 profiles have no human citation of that compound, and those rows carry null_flag true.

compounds.jsoncompounds.csv

License

This dataset is licensed under Creative Commons Attribution 4.0 International (CC BY 4.0). You may copy, redistribute and adapt it, including for commercial use, if you credit The Longevity Archive and link to https://thelongevityarchive.com/data. The license text is at creativecommons.org/licenses/by/4.0.

Cite the compound page for any clinical figure. This dump is a finding aid. It does not replace the study, and it does not carry first-hand reports, prices, seller names or anyone's personal data.

Field dictionary

Version 1 is stable. A later version will change the path or the version field rather than silently rename a key.

slug
The stable URL key. The profile lives at /compounds/{slug} and that page is the cite target.
name
The compound name as it appears on the profile.
class
The catalogue category (Metabolic, Longevity, Healing and Recovery, and the rest).
kind
What kind of thing it is when that is known: Peptide, Small molecule, Supplement, Hormone, Biologic, Blend or Procedure. Null when no kind has been attached.
evidence_band
One word computed from completed trials and counted people, by the published rule on /methodology. Not a verdict on whether the compound works. The arithmetic that produced it is on the profile.
human_study_count
Human-typed citations about this compound, excluding studies labelled as being about a different molecule and excluding retracted records, counted once per PubMed ID or NCT number (or per URL when neither is present). Same definition as the site-wide Human studies count, applied to one profile.
people_n
How many people the completed administered trials on the page state were given the compound. Null when there is no such trial. Zero when those trials exist but none of the lines state a headcount. This undercounts by design.
null_flag
True when the profile has no countable human citation of this compound. That is the same rule as the site-wide Null profiles count.
pmids
PubMed IDs parsed from citation URLs on the profile. Sorted, unique. Includes identifiers that sit on retracted or about-another lines, because those records are still on the page.
ncts
ClinicalTrials.gov NCT numbers parsed from citation URLs on the profile. Sorted, unique.
last_reviewed
Year of the newest cited study on the profile. Null when no citation carries a year. This is a fact about the evidence, not a claim that a person reviewed the page that year.
url
The compound profile on this site. Quote the study, not the summary, for any clinical figure.

Public API

Two GET routes, same JSON the dump uses, CORS open, no key. Unknown slugs return 404 with an error object. These URLs are the contract; do not scrape HTML for the same numbers.

Sample curl

curl https://thelongevityarchive.com/api/compounds
curl https://thelongevityarchive.com/api/compounds/bpc-157
curl https://thelongevityarchive.com/data/compounds.json
curl https://thelongevityarchive.com/data/compounds.csv

How the counts are produced lives on /methodology. Who writes the pages lives on /about.