A package that automatically installs and wraps Hasura CLI binary in isolated manner

Usage no npm install needed!

<script type="module">
  import aaronhayesHasuraCli from 'https://cdn.skypack.dev/@aaronhayes/hasura-cli';




This is a fork from jjangga0214, I maintain this repo to release new versions faster. I often release faster, but sometimes Jjangga0214 does.

An npm package that automatically installs and wraps Hasura CLI binary in isolated manner

license test code style:airbnb code style:prettier .nvmrc yarn:required Conventional Commits Commitizen friendly pr welcome


The Original Hasura CLI, which is not this package, is a compiled binary originally written in go. But just installing it on your system could cause some problems.

  1. Difficult to use different hasura versions on multiple projects.
  2. Inconvenient to ensure every colleagues having same version installed.
  3. Manual installation not specified as npm devDependency.

hasura-cli solves them. It automatically downloads the CLI and exposes the command hasura. Downloaded CLI would be isolated, making it only dedicated to the "project" that installed it. Of course, you can install it as global package as well.


You can just simply install hasura-cli through npm or yarn. Note that this package follows version of the Original Hasura CLI. If you want to check its releases, go here.

Currently there are 2 npm tags (npm tags are different from versions), latest and beta. latest tag refers to Hasura's latest stable version(e.g. v1.1.1 as of writing), while beta, beta version(e.g. v1.2.0-beta.3 as of writing).

Of course, you can install it globally,

npm install --global @aaronhayes/hasura-cli[@tag|@version]

or in a project.

# latest version from latest tag
npm install --save-dev @aaronhayes/hasura-cli@latest

# latest version from beta tag
npm install --save-dev @aaronhayes/hasura-cli@beta

# or specific version
npm install --save-dev @aaronhayes/hasura-cli@1.0.0

# latest version regardless of tag
npm install --save-dev @aaronhayes/hasura-cli

Then you will be able to run hasura command.

For example,

# print hasura version
npx hasura version

Or configure npm scripts on package.json in the way you want. (tip. provide env vars like $HASURA_GRAPHQL_ENDPOINT or $HASURA_GRAPHQL_ADMIN_SECRET)

  "scripts": {
    "hasura": "hasura --project hasura --skip-update-check",
    "hasura:console": "npm run hasura console",
    "hasura:apply": "npm run hasura migrate apply"


Generally, it works on 64 bits architecture of any Linux, macOS, and Windows with node@>=8.

Development (Contribution)


Please read NOTE.md, before getting started.

Environment variables

Environment variables are intended to be only used on development environment.

First, create .env file, and configure it as you want.

cp .env.example .env

You can simply populate the variables by executing yarn dev or yarn dev:no-respawn. Otherwise, you have to manually feed them (e.g. dotenv -- <your command>). That's because this project doesn't use dotenv, but dotenv-cli. So, the application does not read .env by itself.


Whether src/index.ts would install the cli. You can set it false to prevent unwanted downloads.


A directory where Hasura CLI should be installed.


A file name of Hasura CLI.

Getting started

Install dependencies. Lifecycle script postinstall is only for clients who want to install the binary. So, ignore it with --ignore-scripts option. It should also be used on CI.

yarn install --ignore-scripts

On development, you can run

yarn dev
# or
yarn dev:no-respawn
# or
yarn dev:build

yarn dev watches source code and restarts a process when file changes. It does not write compiled js to the file system. ts-node-dev does watching, compiling and restarting.

yarn dev:no-respawn does the same thing except it does not restart.

yarn dev:build logically does the identical job at the high viewpoint. But it compiles (tsc -w) ts, writes js on file system, and run (nodemon) js. concurrently runs tsc and nodemon simualtaneously.

To manually test compiled js, you can run

yarn build # compiles ts to js
yarn start # runs dist/index.js

Other scripts

yarn test # runs all tests (against "*.test.ts")
yarn test:coverage # runs all tests and measures coverage
yarn lint # lint
yarn format # format(fix)

How does this work?

Here is a brief file system tree.

├── dist // to be generated by build process (e.g. `yarn build`), and ignored by git
├── hasura
├── package.json
└── src
    ├── asset.ts
    ├── index.ts
    └── install.ts

package.json exposes the command hasura as a symlink to the flie hasura. Only the directory dist and file hasura are packed as a package.

  "bin": {
    "hasura": "./hasura"
  "files": ["dist", "hasura"]

However, when publishing (npm publish on development environment) the package, the file hasura is just a dummy 'text' file, not a binary flie. The file will be replaced to a binary only when a client installs the package on Linux or MacOS. On windows, unlike linux and macOS, the file hasura is to be removed, and a new file hasura.exe will be created. postinstall lifecycle hook executes dist/index.js, which would install the platform-specfic binary.

The binaries are hosted on GitHub as release assets. src/asset.ts exposes functions of "getting GitHub asset url" and "downloading the asset from the url". src/install.ts exposes a function of "composing them and handling how installation should be processed". src/index.ts uses the function to actually install the asset with some additional control.


MIT License. Copyright © 2019, GIL B. Chan <bnbcmindnpass@gmail.com>