Programmatic generation

Use the generate function when schema generation belongs inside a custom build script. This is useful for monorepos, generated SDK packages, or backend demos that create a client from a freshly built schema.

Generate from a schema string


import { readFile } from 'node:fs/promises';
import { join } from 'node:path';
import { generate } from '@gqlts/cli';
 
await generate({
  schema: await readFile(join(import.meta.dirname, 'schema.graphql'), 'utf8'),
  output: join(import.meta.dirname, 'generated'),
  scalarTypes: {
    DateTime: 'string',
    Upload: 'File',
  },
});

Generate from an endpoint


import { generate } from '@gqlts/cli';
 
await generate({
  endpoint: 'https://api.example.com/graphql',
  output: './generated',
  headers: {
    authorization: `Bearer ${process.env.GRAPHQL_TOKEN}`,
  },
  sortProperties: true,
});

API shape


generate(config: {
  endpoint?: string;
  useGet?: boolean;
  schema?: string;
  output?: string;
  headers?: Record<string, string>;
  scalarTypes?: Record<string, string>;
  onlyEsModules?: boolean;
  onlyCJSModules?: boolean;
  sortProperties?: boolean;
  verbose?: boolean;
}): Promise<void>;

When to prefer the CLI

Prefer the CLI when a package script is enough. Prefer generate(config) when another script already knows where the schema is, which headers to use, or which output package should be rebuilt.