APIs.io indexes 8,968 providers and 32,188 APIs, but an API is never just one file. Behind every provider page is a stack of machine-readable artifacts — the OpenAPI that describes the surface, the Arazzo that sequences the calls, the MCP server that hands it to an agent, the Spectral rules that govern it, the pricing plan that gates it. We index each of those as a first-class specification, with its own browsable section and its own dedicated page for every provider that ships one.
Here is every artifact specification you can browse on APIs.io today, grouped by the job it does.
How the API is described
These are the contract formats — the specifications that tell you (and your tooling) what an API actually offers.
- OpenAPI — the REST/HTTP contract. This is the backbone of the catalog: 32,188 API descriptions across the network, each rendered into interactive reference documentation.
- AsyncAPI — the event-driven counterpart to OpenAPI, describing message-based, streaming, and pub/sub interfaces. 516 event specifications indexed.
- GraphQL — schema definition language (SDL) for GraphQL endpoints, with 978 schemas parsed into browsable types and fields.
- JSON Schema — the data shapes underneath every API. We extract and index 119,624 schemas so you can search the models, not just the operations.
- JSON Structure — the newer, stricter structural schema format, with 52,329 structures indexed.
- Example Payloads — the real request and response bodies that make a spec concrete. 59,034 examples across the catalog.
How agents consume the API
The specifications that turn an API into something an AI agent can pick up and use.
- MCP Servers — Model Context Protocol servers that expose an API’s operations as tools an agent can call. 320 MCP servers indexed and growing.
- Agent Skills — packaged, ready-to-drop skills that ground an agent in a specific provider’s API. 1,489 skills published across providers.
How the calls are sequenced
Single operations are rarely the whole job. These specifications capture the multi-step reality of using an API.
- Arazzo Workflows — the OpenAPI Initiative’s workflow specification, describing ordered, multi-call sequences (sign up → authenticate → make the first real call). 4,563 workflows indexed.
- Collections — Postman and open collection formats, the runnable request bundles teams already share. 9,680 collections in the catalog.
How the API is governed and given meaning
- Governance Rules — Spectral rulesets that lint and govern API descriptions. 3,317 rule artifacts you can browse, fork, and point at your own specs.
- Semantic Vocabularies — JSON-LD contexts that attach shared meaning to API data. 5,435 semantic vocabularies indexed.
The commercial and operational contract
An API’s specification doesn’t stop at the technical surface — how it’s priced and throttled is part of the contract too.
- Pricing Plans — structured plan and tier definitions for 7,123 providers.
- Rate Limits — the throughput ceilings, indexed for 7,086 providers.
- FinOps — the cost and billing metadata that ties consumption to spend, across 7,019 providers.
One index ties them together
None of these live in isolation. Every artifact above is declared in the provider’s APIs.json index — the 0.21 spec treats all of them as typed, url-shaped entries, so a single machine-readable file points at the OpenAPI, the AsyncAPI, the Arazzo, the MCP server, the rules, and the plans for a given provider. Pull one provider’s APIs.json and you have the map to every artifact we’ve indexed for them.
That’s the whole point of APIs.io: not a directory you read with your eyes, but a directory you read with code. Every section above is a browsable site and a queryable surface. Ask the MCP server or the REST API for a provider’s artifacts and you get back the same specifications, ready to feed straight into your own tooling:
curl "https://apis.io/api/v1/search?q=payments&limit=3"
If you publish any of these artifacts and want them indexed, submit your API — the pipeline discovers the specifications you ship and gives each one its own page.