Standards
Editorial policy
A directory that publishes thousands of automatically aggregated pages has an obligation to be explicit about which of its words are data and which are judgement. This page draws that line.
Two kinds of page, two standards
A record page is a factual listing: it repeats what the source registry states and links out so you can check it. An editorial page — methodology, topic overviews, comparisons — makes claims, and is held to the rules below. The distinction is deliberate: we do not dress a listing up as advice.
Claims must be checkable
Any number published on this site is either computed from the build snapshot (and therefore reproducible from the same source the site uses) or attributed to the source it came from. Where a value is unknown, the page says unknown. We do not estimate, and we do not convert "no data" into a confident-sounding generalisation.
No unmeasured performance claims
We do not state that an API is fast, reliable, cheap, easy, production-ready or "the best" unless we have measured it or the operator of that API has published the figure and we are quoting it. Editorial comparisons describe what a record does, what it requires, and what its documented limits are.
Product descriptions are sanitised, never written
Descriptions and taglines for third-party software come from the source registries. We strip markup, normalise whitespace and truncate; we do not paraphrase, generate or embellish. If a record's description is too thin to be useful, the record gets no page of its own rather than being padded with copy of our own.
No paid or sponsored placement
There is no sponsored slot, no paid inclusion, no affiliate ordering and no preferential ranking for any record. Ordering within a topic is by content completeness and then alphabetically. If this ever changes, it will be labelled on the page it affects.
Corrections are applied, not hidden
When a factual error is found, the page is corrected and the correction recorded in the changelog where it changed a published claim. Corrections are not made silently, and pages are not deleted to avoid acknowledging an error.
Authorship and accountability
Content is published by the operator named on the about page, and that organisation is accountable for it. We do not attach invented bylines, fictional credentials or fake reviewer identities to pages. Where a page would benefit from a named human expert, that is a gap we would rather declare than fabricate — and it is listed as open work rather than dressed up with a stock persona.
Never published here
- AI-rewritten descriptions of third-party software presented as fact.
- Comparisons that conclude a winner without stating the criteria and the evidence.
- Truncated previews of the previous page with a new title, or synonym-spun topic variants.
- Popularity or "trending" rankings derived from data that does not contain a popularity signal.
- Reviews of software we have not examined, or hands-on claims about tools we have not run.
Report a problem
If a page here is wrong, thin, misleading or duplicated, report it. Corrections that originate upstream are pursued with the source registry; anything API Atlas normalised incorrectly is fixed directly. See the contact route on the about page.
Related: methodology · changelog · FAQ