tinyRPC

Minimal, end-to-end type-safe RPC framework. Inspired by tRPC v10. Built for speed, developer experience, and production-grade applications.

GitHub • Organization • Twitter

Quick Start

# Clone and setup
git clone https://github.com/renderhq/tinyRPC.git
cd tinyRPC
pnpm install

# Build and run
pnpm build
pnpm dev:server # Terminal 1
pnpm dev:client # Terminal 2

Core Principles

• Type Safety: Automatic type inference from server to client without code generation.
• Protocol Agnostic: Works over HTTP/S, WebSockets, or entirely in-process (Server Callers).
• Edge Ready: Built-in support for Web Standard Request/Response (Fetch API).
• Composable: Powerful middleware system with context transformation and metadata.
• Reactive: Native support for Observables and real-time subscriptions.

Architecture

This is a pnpm monorepo designed for modularity:

• @tinyrpc/server: Procedure builders, router logic, and adapters (Node/Fetch/WS).
• @tinyrpc/client: Generic proxy client, link system, and observable utilities.
• example: A reference implementation featuring infinite loading, file uploads, and rate limiting.

Features and Examples

Internal Server Caller

Execute procedures server-side (e.g., in background tasks) with full type safety and zero network overhead.

const createCaller = createCallerFactory(appRouter);
const caller = createCaller({ user: systemUser, db });

const message = await caller.chat.sendMessage.mutate({ 
  text: 'System update available.' 
});

Data Transformers

Transparently support complex types like Date, Map, Set, or BigInt across the network.

// Server & Client
const t = initTRPC.create({ 
  transformer: {
    serialize: (v) => superjson.serialize(v),
    deserialize: (v) => superjson.deserialize(v),
  }
});

Edge Support

Deploy anywhere using the universal fetchRequestHandler.

export default {
  async fetch(request) {
    return fetchRequestHandler({
      router: appRouter,
      req: request,
      endpoint: '/trpc',
    });
  }
}

Roadmap

We are approaching v1.0. Our focus is on:

• Batching Optimization: Smart request windowing.
• Standardization: Stable API for custom links and adapters.
• Observability: Built-in OpenTelemetry support.

See V2 Proposal for future architectural plans.