api.mdby SuperMD

// api.md

OpenAPI spec → LLM-readable docs.

Paste an OpenAPI 3.x or Swagger 2.x JSON spec. Get clean markdown — endpoints, parameters, schemas, and responses — without the JSON noise. Feed it to any LLM as API context.

OpenAPI JSON
JSON only. Have a YAML spec? Convert it first at yaml-to-json.com or paste it into Claude / ChatGPT and ask for JSON output.
API docs

Docs appear here

Paste an OpenAPI JSON spec on the left

// why not just paste the spec

OpenAPI JSON is noisy. LLMs don't need all of it.

A typical OpenAPI spec is 50–90% structural overhead — deeply nested JSON, repeated $ref chains, example payloads, server variable definitions, and extension keys. The LLM needs the endpoints, the parameters, and the schemas. Nothing else.

This tool resolves $refreferences inline, flattens the schema hierarchy to a readable depth, groups endpoints by tag, and removes everything the LLM doesn't need to generate correct API calls.

The result is typically 40–70% fewer tokens than the raw JSON spec — while containing all the information an LLM needs to write client code, generate tests, or answer questions about the API.

EndpointsMethod, path, and summary for every operation
ParametersName, location, type, required flag, and description as a table
Request schemas$refs resolved inline — no separate schema lookup needed
Response schemasStatus code, description, and response body structure
Tag groupingEndpoints organized by tag for easy navigation
Token savings40–70% smaller than the raw JSON spec

// faq

Frequently asked questions

Which spec versions are supported?

OpenAPI 3.x (3.0, 3.1) and Swagger 2.x. Both JSON formats are supported. YAML is not supported — convert to JSON first using any free online converter or by asking an LLM.

Are $ref schemas resolved?

Yes. References to components/schemas (OpenAPI 3.x) and definitions (Swagger 2.x) are resolved inline up to 3 levels deep. Circular references are truncated with { ... } to prevent infinite loops.

What gets omitted?

Examples, extensions (x-* fields), server variable definitions, security scheme details, external docs links, and deeply nested schema levels beyond 3. These rarely affect the LLM's ability to generate correct API calls.

How should I use the output?

Paste api.md into your LLM context at the start of a session before asking it to write API calls, generate SDK clients, build integrations, or answer questions about the API. It gives the model accurate endpoint and schema knowledge without guessing.