Skip to content



One of the limitations of implementing native add-on modules is that at some point they need to be compiled and linked. In the absence of a downloadable binary, each user of a native add-on will need to compile and link the module. This requires each user to have the necessary C/C++ build tools installed.

An alternative is for the native add-on maintainer to pre-build binaries for supported platforms and architectures and to upload these binaries to a location where users can download them.

This is the specific solution offered by node-pre-gyp.

Note that Node-API support was added to node-pre-gyp in version 0.8.0.

prebuild is an alternative tool that addresses this same issue.

This page describes the modifications required to an Node-API add-on module in order for it to support node-pre-gyp.

Amazon S3

By default, node-pre-gyp uploads generated binaries to the Amazon Web Services (AWS) S3 service.

The separate node-pre-gyp-github module implements publishing binaries to GitHub. Its use is beyond the scope of this tutorial.

Amazon S3 Requirements

Three things are required before uploading binaries to Amazon S3:

  1. An Amazon Web Services account.
  2. An AWS login that permits uploading to Amazon S3.
  3. An Amazon S3 bucket to which the AWS login is permitted to upload objects.

Creating these items is covered on the Amazon Web Services pages.

AWS Credentials

For security reasons, login credentials, such as those for AWS, must never be stored in repositories. For development purposes, node-pre-gyp offers a couple of different ways to store the AWS credentials outside the repository as described in more detail here:

  1. Create a ~/.node_pre_gyprc file with the following contents:

    "accessKeyId": "xxx",
    "secretAccessKey": "xxx"
  2. Specify environment variables like this:

    export node_pre_gyp_accessKeyId=xxx
    export node_pre_gyp_secretAccessKey=xxx

The node-pre-gyp documentation describes additional strategies that may be more appropriate for your CI and automated build environments.

More Information

The node-pre-gyp documentation has complete information about configuring your environments for Amazon S3.


The dependencies and devDependencies properties

Any module using node-pre-gyp obviously has a dependency upon node-pre-gyp. In addition, aws-sdk is required as a devDependency since it’s the code used to upload binaries to Amazon S3.

"dependencies"  : {
  "node-pre-gyp": "0.10.x"
"devDependencies": {
  "aws-sdk": "2.x"

The scripts properties

The scripts install property should specify node-pre-gyp install. The --fallback-to-build argument instructs node-pre-gyp to build the binary on the client machine if a suitable binary cannot be located to download.

"scripts": {
    "install": "node-pre-gyp install --fallback-to-build"

The binary property

The binary property specifies which versions of Node-API your native add-on supports. It also instructs node-pre-gyp where your binaries are located.

"binary": {
    "module_name": "your_module",
    "module_path": "./lib/binding/napi-v{napi_build_version}",
    "remote_path": "./{module_name}/v{version}/{configuration}/",
    "package_name": "{platform}-{arch}-napi-v{napi_build_version}.tar.gz",
    "host": "",
    "napi_versions": [1,3]

Set the module_name value for your project to a valid C variable name.

The sample above shows recommended values for the module_path, remote_path, package_name properties. Set the appropriate bucket name and AWS region values for your project in the host property.

The napi_versions property instructs node-pre-gyp to request one or more Node-API builds. It is required for Node-API add-on modules. For Node-API modules that do not require a specific Node-API version, the recommended value is 3. If your module requires specific Node-API versions, include them in the napi_versions array.

The node-pre-gyp documentation has complete information including Node-API considerations.


New target

A new target must be added to the existing binding.gyp file to copy the binary built by node-gyp into the location specified above by module_path.

  "target_name": "action_after_build",
  "type": "none",
  "dependencies": [ "<(module_name)" ],
  "copies": [
	  "files": [ "<(PRODUCT_DIR)/<(module_name).node" ],
	  "destination": "<(module_path)"


The Node-API header files configure themselves based on the C/C++ symbol NAPI_VERSION which can be communicated by node-pre-gyp to your C/C++ code by including the following property in the original target, typically the first one, in your binding.gyp file:

"defines": [

JavaScript updates

JavaScript code that requires the native code must be updated to dynamically locate the .node file.

var binary = require('node-pre-gyp');
var path = require('path');
var binding_path = binary.find(path.resolve(path.join(__dirname,'./package.json')));
var binding = require(binding_path);


Once these changes are made, you can request node-pre-gyp to build your code:

npm install --build-from-source

If you have tests configured for your project, you can also run those:

npm test

Package and publish

The following two commands will package and publish your native add-on:

./node_modules/.bin/node-pre-gyp package
./node_modules/.bin/node-pre-gyp publish

At this point your binaries are uploaded to Amazon S3 and ready for download.

CI and automated builds

The node-pre-gyp documentation describes in detail how to configure your module for Travis CI (Linux, macOS, and iOS) and AppVeyor (Windows). This gives you the ability to build, test, and publish binaries for system platforms and architectures that you may not otherwise have access to.


node-pre-gyp GitHub project page

Amazon Web Services