Skip to content

Publishing a Rust CLI on npm

Created Aug 10, 2023 – Last Updated Aug 10, 2023

Digital Garden

While building my CLI thanks-contributors (this little script accesses GitHub’s API to get all contributors and their PRs between two distinct points in the history of commits) I decided that I wanted to make this CLI available to my colleagues. And they shouldn’t need to install any additional toolchains besides what they already use on a daily basis (which is Node.js among other things). So it only made sense to publish the CLI to npm so that it can be run with npx @lekoarts/thanks-contributors.

Well, at first this was easier said than done. While searching for “Rust CLI npm” or “Rust binary npm” I’ve found articles talking about binary-install and wasm-pack. They are certainly great projects and good options but I wanted something simpler.

Luckily I found napi-rs. It’s a framework for building pre-compiled Node.js addons in Rust.

#Setting Up napi-rs

I recommend using @napi-rs/cli to initialize your project. The napi-rs getting started guide is a great resource for all options and a deep dive on how it works.

Install the CLI globally:

npm install -g @napi-rs/cli

Afterwards, navigate to your desired parent directory and run the CLI. It’ll create a new folder containing the project:

napi new

When you’re asked for the package name, I recommend using a npm scope because for each platform a separate package will be published. This can trigger spam detection.

In one of the next steps you’re asked for the targets you want to support. There is a short overview page in the napi-rs documentation but this one took me longer to figure out initially. You’ll need to research where your CLI should be able to run. I’ve chosen these options and so far I didn’t get any complaints: darwin-arm64, darwin-x64, linux-arm-gnueabihf, linux-arm64-gnu, linux-arm64-musl, linux-x64-gnu, linux-x64-musl, win32-x64-msvc. So I’m supporting Intel and Apple Silicon Macs, a couple of Linux flavors (including ARM), and Windows.

I recommend enabling GitHub actions as they’ll enable you to automatically build and publish new versions in CI.

Follow the rest of the getting started instructions and also check out the example package walkthrough to learn how to use the scaffolded project.

#Defining an Executable

So you’ve successfully initialized your project and wrote your Rust CLI, e.g. with clap. Great! Now you need to wire up your compiled file in such a way that anyone can run it as a Node.js executable.

Give your exported function inside src/ a suitable name, for example run. When you run npm run build napi-rs will compile your native packages and create a new file at the root called index.js. When you inspect the file you’ll see that it’s quite long — it needs to figure out which package/native binary it should load for your platform. At the very bottom you’ll see run exported as a named function:

const { run } = nativeBinding = run

Which means that you can import run from index.js.

You can use the bin property inside package.json to define one or more executables files. But you can’t directly execute run from index.js as you need to define a small helper file.

At the root of your project, create a bin.js file with the following contents:

#!/usr/bin/env node
const cli = require("./index")
const arguments = process.argv.slice(2) => {

The first line is a shebang and necessary to run the Node.js executable. Since you only want to pass the arguments to the run function, you need to grab them from process.argv. Feel free to console.log(process.argv) to see what it all contains when you run node bin.js.

Last but not least, add the bin property to your root package.json:

"bin": "bin.js"

The next time you publish your package to npm one can directly use the CLI you wrote in Rust 🎉 As mentioned in the beginning, if you need a real-world example of this pattern look at the source code of thanks-contributors.

Want to learn more? Browse my Digital Garden