Quickstart

tRPC combines concepts from REST and GraphQL. If you are unfamiliar with either, take a look at the key Concepts.

Installation

tRPC is split between several packages, so you can install only what you need. Make sure to install the packages you want in the proper sections of your codebase. For this quickstart guide we'll keep it simple and use the vanilla client only. For framework guides, checkout usage with React and usage with Next.js.

Requirements

• tRPC requires TypeScript >= 4.7.0
• We strongly recommend you using "strict": true in your tsconfig.json as we don't officially support non-strict mode.

Start off by installing the @trpc/server and @trpc/client packages:

npm

npm install @trpc/server@next @trpc/client@next

yarn

yarn add @trpc/server@next @trpc/client@next

pnpm

pnpm add @trpc/server@next @trpc/client@next

bun

bun add @trpc/server@next @trpc/client@next

Defining a backend router

Let's walk through the steps of building a typesafe API with tRPC. To start, this API will contain three endpoints with these TypeScript signatures:

ts
type User = { id: string; name: string; };
userList: () => User[];
userById: (id: string) => User;
userCreate: (data: { name: string }) => User;

ts
type User = { id: string; name: string; };
userList: () => User[];
userById: (id: string) => User;
userCreate: (data: { name: string }) => User;

1. Create a router instance

First, let's initialize the tRPC backend. It's good convention to do this in a separate file and export reusable helper functions instead of the entire tRPC object.

server/trpc.ts
ts
import { initTRPC } from '@trpc/server';
 
/**
 * Initialization of tRPC backend
 * Should be done only once per backend!
 */
const t = initTRPC.create();
 
/**
 * Export reusable router and procedure helpers
 * that can be used throughout the router
 */
export const router = t.router;
export const publicProcedure = t.procedure;

server/trpc.ts
ts
import { initTRPC } from '@trpc/server';
 
/**
 * Initialization of tRPC backend
 * Should be done only once per backend!
 */
const t = initTRPC.create();
 
/**
 * Export reusable router and procedure helpers
 * that can be used throughout the router
 */
export const router = t.router;
export const publicProcedure = t.procedure;

Next, we'll initialize our main router instance, commonly referred to as appRouter, in which we'll later add procedures to. Lastly, we need to export the type of the router which we'll later use on the client side.

server/index.ts
ts
import { router } from './trpc';
 
const appRouter = router({
  // ...
});
 
// Export type router type signature,
// NOT the router itself.
export type AppRouter = typeof appRouter;

server/index.ts
ts
import { router } from './trpc';
 
const appRouter = router({
  // ...
});
 
// Export type router type signature,
// NOT the router itself.
export type AppRouter = typeof appRouter;

2. Add a query procedure

Use publicProcedure.query() to add a query procedure to the router.

The following creates a query procedure called userList that returns a list of users from our database:

server/index.ts
ts
import { db } from './db';
import { publicProcedure, router } from './trpc';
 
const appRouter = router({
  userList: publicProcedure
    .query(async () => {
      // Retrieve users from a datasource, this is an imaginary database
      const users = await db.user.findMany();
             
const users: User[]
      return users;
    }),
});

server/index.ts
ts
import { db } from './db';
import { publicProcedure, router } from './trpc';
 
const appRouter = router({
  userList: publicProcedure
    .query(async () => {
      // Retrieve users from a datasource, this is an imaginary database
      const users = await db.user.findMany();
             
const users: User[]
      return users;
    }),
});

3. Using input parser to validate procedure inputs

To implement the userById procedure, we need to accept input from the client. tRPC lets you define input parsers to validate and parse the input. You can define your own input parser or use a validation library of your choice, like zod, yup, or superstruct.

You define your input parser on publicProcedure.input(), which can then be accessed on the resolver function as shown below:

Vanilla

The input parser should be a function that validates and casts the input of this procedure. It should return a strongly typed value when the input is valid or throw an error if the input is invalid.

server/index.ts
ts
const appRouter = router({
  // ...
  userById: publicProcedure
    // The input is unknown at this time. A client could have sent
    // us anything so we won't assume a certain data type.
    .input((val: unknown) => {
      // If the value is of type string, return it.
      // It will now be inferred as a string.
      if (typeof val === 'string') return val;
 
      // Uh oh, looks like that input wasn't a string.
      // We will throw an error instead of running the procedure.
      throw new Error(`Invalid input: ${typeof val}`);
    })
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

server/index.ts
ts
const appRouter = router({
  // ...
  userById: publicProcedure
    // The input is unknown at this time. A client could have sent
    // us anything so we won't assume a certain data type.
    .input((val: unknown) => {
      // If the value is of type string, return it.
      // It will now be inferred as a string.
      if (typeof val === 'string') return val;
 
      // Uh oh, looks like that input wasn't a string.
      // We will throw an error instead of running the procedure.
      throw new Error(`Invalid input: ${typeof val}`);
    })
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

Zod

The input parser can be any ZodType, e.g. z.string() or z.object().

server.ts
ts
import { z } from 'zod';
 
const appRouter = router({
  // ...
  userById: publicProcedure
    .input(z.string())
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

server.ts
ts
import { z } from 'zod';
 
const appRouter = router({
  // ...
  userById: publicProcedure
    .input(z.string())
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

Yup

The input parser can be any YupSchema, e.g. yup.string() or yup.object().

server.ts
ts
import * as yup from 'yup';
 
const appRouter = router({
  // ...
  userById: publicProcedure
    .input(yup.string().required())
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

server.ts
ts
import * as yup from 'yup';
 
const appRouter = router({
  // ...
  userById: publicProcedure
    .input(yup.string().required())
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

Valibot

The input parser can be any Valibot schema wrapped with v.parser or v.parserAsync.

server.ts
ts
import * as v from 'valibot';
 
const appRouter = router({
  // ...
  userById: publicProcedure
    .input(v.parser(v.string()))
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

server.ts
ts
import * as v from 'valibot';
 
const appRouter = router({
  // ...
  userById: publicProcedure
    .input(v.parser(v.string()))
    .query(async (opts) => {
      const { input } = opts;
               
const input: string
      // Retrieve the user with the given ID
      const user = await db.user.findById(input);
             
const user: User | undefined
      return user;
    }),
});

info

Throughout the remaining of this documentation, we will use zod as our validation library.

4. Adding a mutation procedure

Similar to GraphQL, tRPC makes a distinction between query and mutation procedures.

The way a procedure works on the server doesn't change much between a query and a mutation. The method name is different, and the way that the client will use this procedure changes - but everything else is the same!

Let's add a userCreate mutation by adding it as a new property on our router object:

server.ts
ts
const appRouter = router({
  // ...
  userCreate: publicProcedure
    .input(z.object({ name: z.string() }))
    .mutation(async (opts) => {
      const { input } = opts;
               
const input: {
    name: string;
}
      // Create a new user in the database
      const user = await db.user.create(input);
             
const user: {
    name: string;
    id: string;
}
      return user;
    }),
});

server.ts
ts
const appRouter = router({
  // ...
  userCreate: publicProcedure
    .input(z.object({ name: z.string() }))
    .mutation(async (opts) => {
      const { input } = opts;
               
const input: {
    name: string;
}
      // Create a new user in the database
      const user = await db.user.create(input);
             
const user: {
    name: string;
    id: string;
}
      return user;
    }),
});

Serving the API

Now that we have defined our router, we can serve it. tRPC has many adapters so you can use any backend framework of your choice. To keep it simple, we'll use the standalone adapter.

server/index.ts
ts
import { createHTTPServer } from '@trpc/server/adapters/standalone';
 
const appRouter = router({
  // ...
});
 
const server = createHTTPServer({
  router: appRouter,
});
 
server.listen(3000);

server/index.ts
ts
import { createHTTPServer } from '@trpc/server/adapters/standalone';
 
const appRouter = router({
  // ...
});
 
const server = createHTTPServer({
  router: appRouter,
});
 
server.listen(3000);

See the full backend code

server/db.ts
ts
type User = { id: string; name: string };
 
// Imaginary database
const users: User[] = [];
export const db = {
  user: {
    findMany: async () => users,
    findById: async (id: string) => users.find((user) => user.id === id),
    create: async (data: { name: string }) => {
      const user = { id: String(users.length + 1), ...data };
      users.push(user);
      return user;
    },
  },
};

server/db.ts
ts
type User = { id: string; name: string };
 
// Imaginary database
const users: User[] = [];
export const db = {
  user: {
    findMany: async () => users,
    findById: async (id: string) => users.find((user) => user.id === id),
    create: async (data: { name: string }) => {
      const user = { id: String(users.length + 1), ...data };
      users.push(user);
      return user;
    },
  },
};

server/trpc.ts
ts
import { initTRPC } from '@trpc/server';
 
const t = initTRPC.create();
 
export const router = t.router;
export const publicProcedure = t.procedure;

server/trpc.ts
ts
import { initTRPC } from '@trpc/server';
 
const t = initTRPC.create();
 
export const router = t.router;
export const publicProcedure = t.procedure;

server/index.ts
ts
import { createHTTPServer } from "@trpc/server/adapters/standalone";
import { z } from "zod";
import { db } from "./db";
import { publicProcedure, router } from "./trpc";
 
const appRouter = router({
  userList: publicProcedure
    .query(async () => {
      const users = await db.user.findMany();
      return users;
    }),
  userById: publicProcedure
    .input(z.string())
    .query(async (opts) => {
      const { input } = opts;
      const user = await db.user.findById(input);
      return user;
    }),
  userCreate: publicProcedure
    .input(z.object({ name: z.string() }))
    .mutation(async (opts) => {
      const { input } = opts;
      const user = await db.user.create(input);
      return user;
    }),
});
 
export type AppRouter = typeof appRouter;
 
const server = createHTTPServer({
  router: appRouter,
});
 
server.listen(3000);

server/index.ts
ts
import { createHTTPServer } from "@trpc/server/adapters/standalone";
import { z } from "zod";
import { db } from "./db";
import { publicProcedure, router } from "./trpc";
 
const appRouter = router({
  userList: publicProcedure
    .query(async () => {
      const users = await db.user.findMany();
      return users;
    }),
  userById: publicProcedure
    .input(z.string())
    .query(async (opts) => {
      const { input } = opts;
      const user = await db.user.findById(input);
      return user;
    }),
  userCreate: publicProcedure
    .input(z.object({ name: z.string() }))
    .mutation(async (opts) => {
      const { input } = opts;
      const user = await db.user.create(input);
      return user;
    }),
});
 
export type AppRouter = typeof appRouter;
 
const server = createHTTPServer({
  router: appRouter,
});
 
server.listen(3000);

Using your new backend on the client

Let's now move to the client-side code and embrace the power of end-to-end typesafety. When we import the AppRouter type for the client to use, we have achieved full typesafety for our system without leaking any implementation details to the client.

1. Setup the tRPC Client

client/index.ts
ts
import { createTRPCClient, httpBatchLink } from '@trpc/client';
import type { AppRouter } from './server';
//     👆 **type-only** import
 
// Pass AppRouter as generic here. 👇 This lets the `trpc` object know
// what procedures are available on the server and their input/output types.
const trpc = createTRPCClient<AppRouter>({
  links: [
    httpBatchLink({
      url: 'http://localhost:3000',
    }),
  ],
});

client/index.ts
ts
import { createTRPCClient, httpBatchLink } from '@trpc/client';
import type { AppRouter } from './server';
//     👆 **type-only** import
 
// Pass AppRouter as generic here. 👇 This lets the `trpc` object know
// what procedures are available on the server and their input/output types.
const trpc = createTRPCClient<AppRouter>({
  links: [
    httpBatchLink({
      url: 'http://localhost:3000',
    }),
  ],
});

Links in tRPC are similar to links in GraphQL, they let us control the data flow before being sent to the server. In the example above, we use the httpBatchLink, which automatically batches up multiple calls into a single HTTP request. For more in-depth usage of links, see the links documentation.

2. Querying & mutating

You now have access to your API procedures on the trpc object. Try it out!

client/index.ts
ts
// Inferred types
const user = await trpc.userById.query('1');
       
const user: {
    name: string;
    id: string;
} | undefined
 
const createdUser = await trpc.userCreate.mutate({ name: 'sachinraja' });
          
const createdUser: {
    name: string;
    id: string;
}

client/index.ts
ts
// Inferred types
const user = await trpc.userById.query('1');
       
const user: {
    name: string;
    id: string;
} | undefined
 
const createdUser = await trpc.userCreate.mutate({ name: 'sachinraja' });
          
const createdUser: {
    name: string;
    id: string;
}

Full autocompletion

You can open up your Intellisense to explore your API on your frontend. You'll find all of your procedure routes waiting for you along with the methods for calling them.

client/index.ts
ts
// Full autocompletion on your routes
trpc.u;
      
userById
userCreate
userList

client/index.ts
ts
// Full autocompletion on your routes
trpc.u;
      
userById
userCreate
userList

Try it out for yourself!

Next steps

tip

We highly encourage you to check out the example apps to learn about how tRPC is installed in your favorite framework.

tip

By default, tRPC will map complex types like Date to their JSON-equivalent (string in the case of Date). If you want to add to retain the integrity of those types, the easiest way to add support for these is to use superjson as a Data Transformer.

tRPC includes more sophisticated client-side tooling designed for React projects and Next.js.

• Usage with Next.js
• Usage with Express (server-side)
• Usage with React (client-side)