JSON to GraphQL Schema Generator

How to use

1. 1Paste a JSON value into the input on the left, or click Load sample JSON to see a worked example.
2. 2Pick the root type name (defaults to Root, converted to PascalCase) and a Query field name (defaults to the camelCased root, or its plural for list roots).
3. 3Choose Permissive or Strict non-null. Permissive keeps every field nullable to match GraphQL convention; Strict marks always-present, never-null fields with !.
4. 4Toggle Detect ID fields and Detect Date and DateTime to control the ID, Date, and DateTime scalar heuristics. Toggle Emit Query type if you want a mountable Query block at the bottom.
5. 5Read the generated SDL in the panel on the right, review the scalars used and any warnings, and copy the whole schema with one click.

About this tool

JSON to GraphQL Schema Generator turns a sample JSON payload (a REST response, a fixture, an export) into a complete GraphQL Schema Definition Language (SDL) document you can drop straight into apollo-server, graphql-yoga, Mercurius, Pothos, or a graphql-tools setup. The generator parses your input with the browser's native JSON.parse and walks the resulting tree to infer scalar types, named object types, list types, and nullability. Scalar inference splits numbers into Int and Float by checking each sample against the GraphQL Int range (signed 32-bit, -2147483648 to 2147483647) and whether any sample has a non-integer value; Booleans map to Boolean; Strings map to String, with two opt-in heuristics. The ID heuristic types any string under a key called id, uuid, *Id, or *_id as the GraphQL ID scalar; the date heuristic types ISO-8601 date strings (YYYY-MM-DD) as Date and ISO-8601 date-time strings (YYYY-MM-DDTHH:MM:SS with an optional offset) as DateTime, emitting matching scalar declarations at the top of the schema for the custom types so downstream tooling can attach resolvers. Object inference promotes every distinct shape into its own named type. Names are derived from the key in PascalCase, with collision suffixes (Address, Address2, Address3) so two structurally different shapes that happen to share a key do not clobber each other. Inside arrays, the element type name is singularized first (posts becomes Post, stories becomes Story, addresses becomes Address) so the SDL reads naturally. Nullability has two modes. Permissive matches the GraphQL convention: every field is nullable unless you say otherwise, which is the safe default for clients. Strict adds the ! non-null marker to fields that were always present and never null in the sample data, but it still keeps observed nulls and missing keys nullable so the schema reflects reality. Mixed-type fields fall back to a JSON scalar with a warning so you can decide on a union or interface manually. Keys that are not valid GraphQL identifiers (kebab-case, leading digits, spaces, dots) are renamed to a safe identifier and the original key is preserved in a SDL comment above the field. The optional Query block emits a single root field named from the root type (camelCased for object roots, pluralized for list roots) so the schema is mountable out of the box; the field name is editable. Everything runs in your browser, no JSON ever leaves the page, and the result includes a list of scalars used, every generated type, and any warnings to review.

Free to use. Works in your browser. No signup, no login.