protobuf.js

protobuf.js is a very fast, conformant, and unusually versatile JavaScript implementation of Google’s Protocol Buffers for Node.js and browsers. It works with .proto files out of the box, does not require the protoc compiler, and supports both runtime reflection and specialized code generation with strong TypeScript declarations.

Beyond basic encode/decode, protobuf.js supports programmatic schema definitions (no .proto file needed), transport-agnostic RPC services, optional extensions, and a companion CLI package (protobufjs-cli) for generating static code, TypeScript typings, and JSON reflection bundles — with an optional protoc-gen-pbjs path for teams that prefer a protoc-based toolchain.

What You Get

• Runtime .proto file loading and reflection-based encode/decode with no protoc dependency
• A protobufjs-cli add-on for generating static JS, TypeScript declarations, and JSON reflection bundles ahead of time
• Programmatic schema definitions for building message types without a .proto file at all
• Transport-agnostic Service/RPC support for wiring protobuf messages to any transport layer
• Lightweight light/minimal build variants for bundle-size-sensitive browser deployments

Common Use Cases

• Encoding/decoding Protocol Buffer messages in a Node.js backend or browser frontend without a native protoc toolchain
• Generating static, tree-shakeable JS/TypeScript bindings from .proto schemas at build time via the CLI
• Implementing gRPC-style or custom binary RPC services in JavaScript using protobuf.js’s transport-agnostic service layer
• Interoperating with Protobuf-based systems written in Go, Java, Python, or C++ from a JS/TS codebase

Under The Hood

Architecture - the library’s core lives in src/ as a set of focused modules: root.js/namespace.js/type.js/field.js implement the reflective schema object model, encoder.js/decoder.js/reader.js/writer.js (under src/) handle wire-format serialization, parse.js/tokenize.js implement a hand-written .proto parser, and rpc/ provides the transport-agnostic Service abstraction; index.js, light.js, and minimal.js expose progressively smaller entrypoints for full/reflection-light/wire-only use cases. The separate cli/ package builds on this core to generate static code and TypeScript declarations ahead of time. Tech Stack - plain JavaScript (CommonJS) with hand-authored .d.ts TypeScript declaration files (index.d.ts, light.d.ts, minimal.d.ts), Node.js-based build/release tooling (release-please, GitHub Actions CI), and a google/ directory bundling well-known .proto types (Any, Timestamp, Struct, etc.). Code Quality - the tests/ directory contains 101 test files covering the parser, wire format, reflection API, and RPC layer, CI runs on every push via GitHub Actions, and the project has been continuously maintained by the community since dcodeIO’s original release, now under org-based stewardship. API Design - protobuf.load()/root.lookupType() gives a low-boilerplate entrypoint for dynamic schema loading, while the CLI’s generated static code gives TypeScript users compile-time-checked message types; the split between index/light/minimal entrypoints lets consumers explicitly trade reflection features for bundle size.