Skip to main content
effect-orpc keeps oRPC’s input and output schema model. Use .input(...) and .output(...) before .effect(...).
schemas.ts
import * as z from "zod";

const GetUserInput = z.object({
  id: z.string().min(1),
});

const User = z.object({
  id: z.string(),
  name: z.string(),
});

export const getUser = effectProcedure
  .input(GetUserInput)
  .output(User)
  .effect(function* ({ input }) {
    return yield* UsersRepo.findById(input.id);
  });

​
Inferred input

The handler sees the parsed schema output as input.
parsed-input.ts
const SearchInput = z.object({
  query: z.string().trim(),
  limit: z.number().int().positive().default(20),
});

const search = effectProcedure.input(SearchInput).effect(function* ({ input }) {
  // input.limit is a number, including the default when omitted.
  return yield* UsersRepo.search(input.query, input.limit);
});

​
Output validation

Use .output(...) for public response contracts.
output.ts
const PublicUser = z.object({
  id: z.string(),
  name: z.string(),
});

const me = effectProcedure.output(PublicUser).effect(function* () {
  const user = yield* CurrentUser;
  return user;
});
Output schemas validate what leaves the procedure. Keep internal fields out of public output schemas unless they are intentionally part of the API.

​
Next step

Provide services with Service injection.
Last modified on June 15, 2026