Read an OpenAPI or Swagger spec without setting up Swagger UI. Paste YAML or JSON, or drop the file, and browse every endpoint grouped by tag — methods, parameters, request bodies and responses with their schemas written out as readable types. References are resolved, and problems like undeclared path parameters, duplicate operationIds and broken $refs are listed. Search to find an endpoint. It's free, runs in your browser, and the spec never leaves your device.
| Concept | OpenAPI 3 | Swagger 2 |
|---|---|---|
| Reusable schemas | components/schemas | definitions |
| Request body | requestBody.content | parameter with in: body |
| Base URL | servers[].url | schemes + host + basePath |
| Response body | responses.*.content.*.schema | responses.*.schema |
Yes — OpenAPI 3.0 and 3.1 and Swagger 2.0, in YAML or JSON.
Internal ones (#/components/… and #/definitions/…) are, including circular ones. References to other files are listed but not fetched.
No. Calling the API from this page would send your data to it and usually fail on CORS. Use curl or an API client for that.
Missing references, undeclared path parameters, duplicate operationIds, external references and missing title or version.
No. It's parsed in your browser.
The viewer parses the specification as JSON or YAML and detects whether it's OpenAPI 3.x or Swagger 2.0, then lists every operation grouped by tag. Path-level parameters are merged into each operation, with operation-level parameters taking precedence; Swagger 2's body parameter and OpenAPI 3's requestBody are both shown as the request body with its content type. Internal references — #/components/... or #/definitions/... — are resolved, with JSON Pointer escaping handled and circular references detected and labeled instead of looping forever. Schemas are summarized as compact type signatures, such as { id: string (uuid), name: string, tag?: string } or Pet[], with optional properties marked and oneOf, anyOf, allOf and enums written out, so a response's shape can be read at a glance. Every $ref in the document is checked, and the tool reports references that don't exist, external file references (which aren't fetched), path parameters used in a URL but not declared, duplicate operationIds and missing info fields. There's deliberately no live request button.