Bun Runtime

The Bun Runtime is designed to start fast and run fast.

Under the hood, Bun uses the JavaScriptCore engine, which is developed by Apple for Safari. In most cases, the startup and running performance is faster than V8, the engine used by Node.js and Chromium-based browsers. Its transpiler and runtime are written in Zig, a modern, high-performance language. On Linux, this translates into startup times 4x faster than Node.js.

This benchmark is based on running a simple Hello World script on Linux

Run a file

Use bun run to execute a source file.

$ bun run index.js

Bun supports TypeScript and JSX out of the box. Every file is transpiled on the fly by Bun's fast native transpiler before being executed.

$ bun run index.js
$ bun run index.jsx
$ bun run index.ts
$ bun run index.tsx

Alternatively, you can omit the run keyword and use the "naked" command; it behaves identically.

$ bun index.tsx
$ bun index.js

--watch

To run a file in watch mode, use the --watch flag.

$ bun --watch run index.tsx

$ bun --watch run dev # ✔️ do this
$ bun run dev --watch # ❌ don't do this

Run a package.json script

$ bun [bun flags] run <script> [script flags]

Your package.json can define a number of named "scripts" that correspond to shell commands.

{
  // ... other fields
  "scripts": {
    "clean": "rm -rf dist && echo 'Done.'",
    "dev": "bun server.ts"
  }
}

Use bun run <script> to execute these scripts.

$ bun run clean
$ rm -rf dist && echo 'Done.'

Cleaning...
Done.

Bun executes the script command in a subshell. On Linux & macOS, it checks for the following shells in order, using the first one it finds: bash, sh, zsh. On Windows, it uses bun shell to support bash-like syntax and many common commands.

Scripts can also be run with the shorter command bun <script>, however if there is a built-in bun command with the same name, the built-in command takes precedence. In this case, use the more explicit bun run <script> command to execute your package script.

$ bun run dev

To see a list of available scripts, run bun run without any arguments.

$ bun run

quickstart scripts:

 bun run clean
   rm -rf dist && echo 'Done.'

 bun run dev
   bun server.ts

2 scripts

Bun respects lifecycle hooks. For instance, bun run clean will execute preclean and postclean, if defined. If the pre<script> fails, Bun will not execute the script itself.

--bun

It's common for package.json scripts to reference locally-installed CLIs like vite or next. These CLIs are often JavaScript files marked with a shebang to indicate that they should be executed with node.

#!/usr/bin/env node

// do stuff

By default, Bun respects this shebang and executes the script with node. However, you can override this behavior with the --bun flag. For Node.js-based CLIs, this will run the CLI with Bun instead of Node.js.

$ bun run --bun vite

Filtering

In monorepos containing multiple packages, you can use the --filter argument to execute scripts in many packages at once.

Use bun run --filter <name_pattern> <script> to execute <script> in all packages whose name matches <name_pattern>. For example, if you have subdirectories containing packages named foo, bar and baz, running

$ bun run --filter 'ba*' <script>

will execute <script> in both bar and baz, but not in foo.

Find more details in the docs page for filter.

bun run - to pipe code from stdin

bun run - lets you read JavaScript, TypeScript, TSX, or JSX from stdin and execute it without writing to a temporary file first.

$ echo "console.log('Hello')" | bun run -

Hello

You can also use bun run - to redirect files into Bun. For example, to run a .js file as if it were a .ts file:

$ echo "console.log!('This is TypeScript!' as any)" > secretly-typescript.js
$ bun run - < secretly-typescript.js

This is TypeScript!

For convenience, all code is treated as TypeScript with JSX support when using bun run -.

bun run --console-depth

Control the depth of object inspection in console output with the --console-depth flag.

$ bun --console-depth 5 run index.tsx

This sets how deeply nested objects are displayed in console.log() output. The default depth is 2. Higher values show more nested properties but may produce verbose output for complex objects.

const nested = { a: { b: { c: { d: "deep" } } } };
console.log(nested);
// With --console-depth 2 (default): { a: { b: [Object] } }
// With --console-depth 4: { a: { b: { c: { d: 'deep' } } } }

bun run --smol

In memory-constrained environments, use the --smol flag to reduce memory usage at a cost to performance.

$ bun --smol run index.tsx

This causes the garbage collector to run more frequently, which can slow down execution. However, it can be useful in environments with limited memory. Bun automatically adjusts the garbage collector's heap size based on the available memory (accounting for cgroups and other memory limits) with and without the --smol flag, so this is mostly useful for cases where you want to make the heap size grow more slowly.

Resolution order

Absolute paths and paths starting with ./ or .\\ are always executed as source files. Unless using bun run, running a file with an allowed extension will prefer the file over a package.json script.

When there is a package.json script and a file with the same name, bun run prioritizes the package.json script. The full resolution order is:

1. package.json scripts, eg bun run build
2. Source files, eg bun run src/main.js
3. Binaries from project packages, eg bun add eslint && bun run eslint
4. (bun run only) System commands, eg bun run ls

CLI Usage

$ bun run <file or script>

General Execution Options

Workspace Management

Runtime & Process Control

Development Workflow

Debugging

Dependency & Module Resolution

Transpilation & Language Features

Networking & Security

Global Configuration & Context

Examples

Run a JavaScript or TypeScript file:

$ bun run ./index.js
$ bun run ./index.tsx

Run a package.json script:

$ bun run dev
$ bun run lint