API ATLAS

Questions and answers

Frequently asked questions

Answers that state the actual numbers and the actual limits, including where the data is incomplete.

Is API Atlas free to use?
Yes. The whole catalogue is public, there is no account, no paywall and no paid inclusion. No record is ranked by payment — ordering within a topic is by content completeness and then alphabetically.
What is the difference between the API and MCP sections?
The API section covers 4,352 conventional HTTP APIs you call from your own code. The MCP section covers 5,809 Model Context Protocol servers, which are installed into an AI client so the model can call tools on your behalf. Same record schema, different integration model, so they are kept on separate routes.
How many records are in the catalogue?
10,161 records at the current snapshot — 4,352 APIs and 5,809 MCP servers, grouped into 53 canonical topics. 4,456 of them are substantial enough to get a dedicated page; the other 5,705 are listed on their topic page with a direct link to their own documentation.
Why do some records have no page of their own?
Because a large share of aggregated records ship a description of one line and nothing else. Generating a page for each would add 5,705 near-empty URLs without adding a useful result. Nothing is hidden by this: every record appears in its topic listing, links straight to its own documentation, and stays searchable in the interactive catalogue — it simply does not get a URL of its own until the source data is good enough to make one worth reading.
Where does the catalogue data come from?
Compiled automatically from public registries: apis.guru, public-apis, Official MCP registry, top-mcps.com. Records are aggregated, then normalised — descriptions sanitised, categories consolidated, schema unified. Descriptions are never rewritten by us.
Why do the categories look different from the source registries?
The sources ship 137 distinct category strings for 10,161 records, including the same idea spelled several ways, free-form buckets and a few malformed labels. They are consolidated into 53 topics so that similar records are not split across competing pages.
What does "authentication unknown" mean on a record?
It means the source registry did not state a scheme. It is not a claim that the record is unauthenticated. 2,477 records are in that state. Always check the record's own documentation before integrating — the record page links to it.
Can I rely on the base URL, pricing or CORS fields?
Treat every field as a copy of what the source registry stated at snapshot time (2026-09-18), not as a live probe. Coverage is uneven — 2,527 of 10,161 records publish a base URL — so a missing field means "not stated", never "not applicable". Verify against the linked documentation.
How is a topic page created?
A topic page is generated when a consolidated topic holds 8 or more records, and those pages are submitted in the sitemap. Below that the topic keeps its page — it stays browsable and linked from the category index — but the page is served as noindex, follow and left out of the sitemap, because a page listing a handful of records has nothing to offer a search engine.
How do I suggest a record or report an error?
Write to hi@zordonintelligence.pl. Because the catalogue is compiled from upstream sources, most factual corrections are fixed at the source — but send it either way and it will be followed through. Corrections that change a published claim are recorded on the changelog page.
How often is the site updated?
Each production build takes a fresh snapshot of the catalogue, and builds reuse a cached copy for up to 12 hours. Site copy and structure change when there is a reason to change them; the changelog records those separately from data refreshes.
Is there an API for the catalogue itself?
Yes. The same public JSON API that builds this site serves the catalogue to anyone, with no account and no key, and it is documented in the API reference: the endpoints, the query parameters and the behaviours worth knowing before you build on it.

Related: API reference · methodology · editorial policy · statistics · about