Cowboy hat
Cowboy hat

Table of Contents


Jake is the JavaScript build tool for NodeJS. Jake has been around since the very early days of Node, and is very full featured and well tested.

Installing with NPM

Install globally with:

npm install -g jake

Or you may also install it as a development dependency in a package.json file:

// package.json
"devDependencies": {
  "jake": "latest"

Then install it with npm install

Note Jake is intended to be mostly a command-line tool, but lately there have been changes to it so it can be either embedded, or run from inside your project.

Windows, installing from source

For Windows users installing from source, there are some additional steps.

Assumed: current directory is the same directory where node.exe is present.

Get Jake:

git clone git:// node_modules/jake

Copy jake.bat and jake to the same directory as node.exe

copy node_modules/jake/jake.bat jake.bat
copy node_modules/jake/jake jake

Add the directory of node.exe to the environment PATH variable.

Basic usage

jake [options ...] [env variables ...] target


Jake is a simple JavaScript build program with capabilities similar to the
regular make or rake command.

Jake has the following features:
    * Jakefiles are in standard JavaScript syntax
    * Tasks with prerequisites
    * Namespaces for tasks
    * Async task execution


--version                   Display the Jake version.

--help                      Display help message.

-f *FILE*
--jakefile *FILE*           Use FILE as the Jakefile.

--directory *DIRECTORY*     Change to DIRECTORY before running tasks.

--quiet                     Do not log messages to standard output.

--jakelibdir *JAKELIBDIR*   Auto-import any .jake files in JAKELIBDIR.
                            (default is 'jakelib')

--always-make               Unconditionally make all targets.

--trace                     Enable full backtrace.

--tasks                     Display the tasks (matching optional PATTERN)
                            with descriptions, then exit.

Jakefile syntax

A Jakefile is just executable JavaScript. You can include whatever JavaScript you want in it.

API Docs

API docs can be found here.


Use task to define tasks. It has one required argument, the task-name, and three optional arguments:

task(name, [prerequisites], [action], [opts]);

The name argument is a String with the name of the task, and prerequisites is an optional Array arg of the list of prerequisite tasks to perform first.

The action is a Function defining the action to take for the task. (Note that Object-literal syntax for name/prerequisites in a single argument a la Rake is also supported, but JavaScript's lack of support for dynamic keys in Object literals makes it not very useful.) The action is invoked with the Task object itself as the execution context (i.e, "this" inside the action references the Task object).

The opts argument is the normal JavaScript-style 'options' object. When a task's operations are asynchronous, the async property should be set to true, and the task must call complete() to signal to Jake that the task is done, and execution can proceed. By default the async property is false.

Tasks created with task are always executed when asked for (or are a prerequisite). Tasks created with file are only executed if no file with the given name exists or if any of its file-prerequisites are more recent than the file named by the task. Also, if any prerequisite is a regular task, the file task will always be executed.

Use desc to add a string description of the task.

Here's an example:

desc('This is the default task.');
task('default', function (params) {
  console.log('This is the default task.');

desc('This task has prerequisites.');
task('hasPrereqs', ['foo', 'bar', 'baz'], function (params) {
  console.log('Ran some prereqs first.');

And here's an example of an asynchronous task:

desc('This is an asynchronous task.');
task('asyncTask', {async: true}, function () {
  setTimeout(complete, 1000);

A Task is also an EventEmitter which emits the 'start' event when it begins to run, and the 'complete' event when it is finished. This allows asynchronous tasks to be run from within other tasks via either invoke or execute, and ensures they will complete before the rest of the containing task executes. See the section "Running tasks from within other tasks," below.


Create a file-task by calling file.

File-tasks create a file from one or more other files. With a file-task, Jake checks both that the file exists, and also that it is not older than the files specified by any prerequisite tasks. File-tasks are particularly useful for compiling something from a tree of source files.

desc('This builds a minified JS file for production.');
file('foo-minified.js', ['bar', 'foo-bar.js', 'foo-baz.js'], function () {
  // Code to concat and minify goes here


Create a directory-task by calling directory.

Directory-tasks create a directory for use with for file-tasks. Jake checks for the existence of the directory, and only creates it if needed.

desc('This creates the bar directory for use with the foo-minified.js file-task.');

This task will create the directory when used as a prerequisite for a file-task, or when run from the command-line.


Use namespace to create a namespace of tasks to perform. Call it with two arguments:

namespace(name, namespaceTasks);

Where is name is the name of the namespace, and namespaceTasks is a function with calls inside it to task or desc definining all the tasks for that namespace.

Here's an example:

desc('This is the default task.');
task('default', function () {
  console.log('This is the default task.');

namespace('foo', function () {
  desc('This the foo:bar task');
  task('bar', function () {
    console.log('doing foo:bar task');

  desc('This the foo:baz task');
  task('baz', ['default', 'foo:bar'], function () {
    console.log('doing foo:baz task');


In this example, the foo:baz task depends on the the default and foo:bar tasks.


When you add a filename as a prerequisite for a task, but there is not a a file-task defined for it, Jake can create file-tasks on the fly from Rules.

Here's an example:

rule('.o', '.c', {async: true}, function () {
  var cmd = 'cc ' + this.source + ' -c -o ' +;
  jake.exec(cmd, function () {

This rule will take effect for any task-name that ends in '.o', but will require the existence of a prerequisite source file with the same name ending in '.c'.

For example, with this rule, if you reference a task 'foobarbaz.o' as a prerequisite somewhere in one of your Jake tasks, rather than complaining about this file not existing, or the lack of a task with that name, Jake will automatically create a FileTask for 'foobarbaz.o' with the action specified in the rule you've defined. (The usual action would be to create 'foobarbaz.o' from 'foobarbaz.c'). If 'foobarbaz.c' does not exist, it will recursively attempt synthesize a viable rule for it as well.

Regex patterns

You can use regular expressions to match file extensions as well:

rule(/\.o$/, '.c', {async: true}, function () {
  var cmd = 'cc ' + this.source + ' -c -o ' +;
  jake.exec(cmd, function () {

Source files from functions

You can also use a function to calculate the name of the desired source-file to use, instead of assuming simple suffix-substitution:

// Match .less.css or .scss.css and run appropriate preprocessor
var getSourceFilename = function (name) {
  // Strip off the extension for the filename
  return name.replace(/\.css$/, '');
rule(/\.\w{2,4}\.css$/, getSourceFilename, {async: true}, function () {
  // Get appropriate preprocessor for this.source, e.g., foo.less
  // Generate a file with filename of, e.g., foo.less.css

Passing parameters to jake

Parameters can be passed to Jake two ways: plain arguments, and environment variables.

To pass positional arguments to the Jake tasks, enclose them in square braces, separated by commas, after the name of the task on the command-line. For example, with the following Jakefile:

desc('This is an awesome task.');
task('awesome', function (a, b, c) {
  console.log(a, b, c);

You could run jake like this:

jake awesome[foo,bar,baz]

And you'd get the following output:

foo bar baz

Note that you cannot uses spaces between the commas separating the parameters.

Any parameters passed after the Jake task that contain an equals sign (=) will be added to process.env.

With the following Jakefile:

desc('This is an awesome task.');
task('awesome', function (a, b, c) {
  console.log(a, b, c);
  console.log(process.env.qux, process.env.frang);

You could run jake like this:

jake awesome[foo,bar,baz] qux=zoobie frang=asdf

And you'd get the following output:

foo bar baz
zoobie asdf

Running jake with no arguments runs the default task.

Note for zsh users : you will need to escape the brackets or wrap in single quotes like this to pass parameters :

jake 'awesome[foo,bar,baz]'

An other solution is to deactivate permanently file-globbing for the jake command. You can do this by adding this line to your .zshrc file :

alias jake="noglob jake"

Cleanup after all tasks run, jake 'complete' event

The base 'jake' object is an EventEmitter, and fires a 'start' event before running, an 'error' event after an uncaught exception, and a 'complete' event after running all tasks.

This is sometimes useful when a task starts a process which keeps the Node event-loop running (e.g., a database connection). If you know you want to stop the running Node process after all tasks have finished, you can set a listener for the 'complete' event, like so:

jake.addListener('complete', function () {

Running tasks from within other tasks

Jake supports the ability to run a task from within another task via the invoke and execute methods.

The invoke method will run the desired task, along with its prerequisites:

desc('Calls the foo:bar task and its prerequisites.');
task('invokeFooBar', function () {
  // Calls foo:bar and its prereqs

The invoke method will only run the task once, even if you call it repeatedly.

desc('Calls the foo:bar task and its prerequisites.');
task('invokeFooBar', function () {
  // Calls foo:bar and its prereqs
  // Does nothing

The execute method will run the desired task without its prerequisites:

desc('Calls the foo:bar task without its prerequisites.');
task('executeFooBar', function () {
  // Calls foo:bar without its prereqs

Calling execute repeatedly will run the desired task repeatedly.

desc('Calls the foo:bar task without its prerequisites.');
task('executeFooBar', function () {
  // Calls foo:bar without its prereqs
  // Can keep running this over and over

If you want to run the task and its prerequisites more than once, you can use invoke with the reenable method.

desc('Calls the foo:bar task and its prerequisites.');
task('invokeFooBar', function () {
  // Calls foo:bar and its prereqs
  // Does nothing
  // Only re-runs foo:bar, but not its prerequisites

The reenable method takes a single Boolean arg, a 'deep' flag, which reenables the task's prerequisites if set to true.

desc('Calls the foo:bar task and its prerequisites.');
task('invokeFooBar', function () {
  // Calls foo:bar and its prereqs
  // Does nothing
  // Re-runs foo:bar and all of its prerequisites

It's easy to pass params on to a sub-task run via invoke or execute:

desc('Passes params on to other tasks.');
task('passParams', function () {
  var t = jake.Task['foo:bar'];
  // Calls foo:bar, passing along current args
  t.invoke.apply(t, arguments);

Getting values out of tasks

Passing a value to the complete function for async tasks (or simply returning a value from sync tasks) will set a 'value' property on the completed task. This same value will also be passed as the task emits its 'complete' event.

After a task is completed, this value will be also available in the '.value' property on the task. Calling reenable on the task will clear this value.

task('environment', {async: true}, function () {
  // Do some sort of I/O to figure out the environment value
  doSomeAsync(function (err, val) {
    if (err) { throw err }

task("someTaskWithEnvViaPrereq", ["envrionment"], function () {
  api = jake.Task["envrionment"].value;

task("someTaskWithEnvViaInvoke", {async: true}, function () {
  var env = jake.Task["envrionment"];
  env.addListener('complete', function (api) {

Managing asynchrony without prereqs (e.g., when using invoke)

You can mix sync and async without problems when using normal prereqs, because the Jake execution loop takes care of the difference for you. But when you call invoke or execute, you have to manage the asynchrony yourself.

Here's a correct working example:

task('async1', ['async2'], {async: true}, function () {
    console.log('-- async1 start ----------------');
    setTimeout(function () {
        console.log('-- async1 done ----------------');
    }, 1000);

task('async2', {async: true}, function () {
    console.log('-- async2 start ----------------');
    setTimeout(function () {
        console.log('-- async2 done ----------------');
    }, 500);

task('init', ['async1', 'async2'], {async: true}, function () {
    console.log('-- init start ----------------');
    setTimeout(function () {
        console.log('-- init done ----------------');
    }, 100);

task('default', {async: true}, function () {
  console.log('-- default start ----------------');
  var init = jake.Task.init;
  init.addListener('complete', function () {
    console.log('-- default done ----------------');

You have to declare the "default" task as asynchronous as well, and call complete on it when "init" finishes. Here's the output:

-- default start ----------------
-- async2 start ----------------
-- async2 done ----------------
-- async1 start ----------------
-- async1 done ----------------
-- init start ----------------
-- init done ----------------
-- default done ----------------

You get what you expect -- "default" starts, the rest runs, and finally "default" finishes.

Evented tasks

Tasks are EventEmitters. They can fire 'complete' and 'error' events.

If a task called via invoke is asynchronous, you can set a listener on the 'complete' event to run any code that depends on it.

desc('Calls the async foo:baz task and its prerequisites.');
task('invokeFooBaz', {async: true}, function () {
  var t = jake.Task['foo:baz'];
  t.addListener('complete', function () {
    console.log('Finished executing foo:baz');
    // Maybe run some other code
    // ...
    // Complete the containing task
  // Kick off foo:baz

If you want to handle the errors in a task in some specific way, you can set a listener for the 'error' event, like so:

namespace('vronk', function () {
  task('groo', function () {
    var t = jake.Task['vronk:zong'];
    t.addListener('error', function (e) {

  task('zong', function () {
    throw new Error('OMFGZONG');

If no specific listener is set for the "error" event, errors are handled by Jake's generic error-handling.

Aborting a task

You can abort a task by calling the fail function, and Jake will abort the currently running task. You can pass a customized error message to fail:

desc('This task fails.');
task('failTask', function () {
  fail('Yikes. Something back happened.');

You can also pass an optional exit status-code to the fail command, like so:

desc('This task fails with an exit-status of 42.');
task('failTaskQuestionCustomStatus', function () {
  fail('What is the answer?', 42);

The process will exit with a status of 42.

Uncaught errors will also abort the currently running task.

Showing the list of tasks

Passing jake the -T or --tasks flag will display the full list of tasks available in a Jakefile, along with their descriptions:

$ jake -T
jake default       # This is the default task.
jake asdf          # This is the asdf task.
jake concat.txt    # File task, concating two files together
jake failure       # Failing task.
jake lookup        # Jake task lookup by name.
jake foo:bar       # This the foo:bar task
jake foo:fonebone  # This the foo:fonebone task

Setting a value for -T/--tasks will filter the list by that value:

$ jake -T foo
jake foo:bar       # This the foo:bar task
jake foo:fonebone  # This the foo:fonebone task

The list displayed will be all tasks whose namespace/name contain the filter-string.

Internally, passing this CLI flag calls jake.showAllTaskDescriptions, and passes it the filter -- so jake -T foo is equivalent to calling jake.showAllTaskDescriptions('foo');.

Breaking things up into multiple files

Jake will automatically look for files with a .jake extension in a 'jakelib' directory in your project, and load them (via require) after loading your Jakefile. (The directory name can be overridden using the -J/--jakelibdir command-line option.)

This allows you to break your tasks up over multiple files -- a good way to do it is one namespace per file: e.g., a zardoz namespace full of tasks in 'jakelib/zardox.jake'.

Note that these .jake files each run in their own module-context, so they don't have access to each others' data. However, the Jake API methods, and the task-hierarchy are globally available, so you can use tasks in any file as prerequisites for tasks in any other, just as if everything were in a single file.

Environment-variables set on the command-line are likewise also naturally available to code in all files via process.env.


Since shelling out in Node is an asynchronous operation, Jake comes with a few useful file-utilities with a synchronous API, that make scripting easier.

The jake.mkdirP utility recursively creates a set of nested directories. It will not throw an error if any of the directories already exists. Here's an example:


The jake.cpR utility does a recursive copy of a file or directory. It takes two arguments, the file/directory to copy, and the destination. Note that this command can only copy files and directories; it does not perform globbing (so arguments like '*.txt' are not possible).

jake.cpR(path.join(sourceDir, '/templates'), currentDir);

This would copy 'templates' (and all its contents) into currentDir.

The jake.readdirR utility gives you a recursive directory listing, giving you output somewhat similar to the Unix find command. It only works with a directory name, and does not perform filtering or globbing.


This would return an array of filepaths for all files in the 'pkg' directory, and all its subdirectories.

The jake.rmRf utility recursively removes a directory and all its contents.


This would remove the 'pkg' directory, and all its contents.

Running shell-commands: jake.exec and jake.createExec

Jake also provides a more general utility function for running a sequence of shell-commands.


The jake.exec command takes an array of shell-command strings, and an optional callback to run after completing them. Here's an example from Jake's Jakefile, that runs the tests:

desc('Runs the Jake tests.');
task('test', {async: true}, function () {
  var cmds = [
    'node ./tests/parseargs.js'
  , 'node ./tests/task_base.js'
  , 'node ./tests/file_task.js'
  jake.exec(cmds, {printStdout: true}, function () {
    console.log('All tests passed.');

desc('Runs some apps in interactive mode.');
task('interactiveTask', {async: true}, function () {
  var cmds = [
    'node' // Node conosle
  , 'vim' // Open Vim
  jake.exec(cmds, {interactive: true}, function () {

It also takes an optional options-object, with the following options:

  • interactive (tasks are interactive, trumps printStdout and printStderr below, default false)

  • printStdout (print to stdout, default false)

  • printStderr (print to stderr, default false)

  • breakOnError (stop execution on error, default true)

This command doesn't pipe input between commands -- it's for simple execution.

jake.createExec and the evented Exec object

Jake also provides an evented interface for running shell commands. Calling jake.createExec returns an instance of jake.Exec, which is an EventEmitter that fires events as it executes commands.

It emits the following events:

  • 'cmdStart': When a new command begins to run. Passes one arg, the command being run.

  • 'cmdEnd': When a command finishes. Passes one arg, the command being run.

  • 'stdout': When the stdout for the child-process recieves data. This streams the stdout data. Passes one arg, the chunk of data.

  • 'stderr': When the stderr for the child-process recieves data. This streams the stderr data. Passes one arg, the chunk of data.

  • 'error': When a shell-command exits with a non-zero status-code. Passes two args -- the error message, and the status code. If you do not set an error handler, and a command exits with an error-code, Jake will throw the unhandled error. If breakOnError is set to true, the Exec object will emit and 'error' event after the first error, and stop any further execution.

To begin running the commands, you have to call the run method on it. It also has an append method for adding new commands to the list of commands to run.

Here's an example:

var ex = jake.createExec(['']);
ex.addListener('error', function (msg, code) {
  if (code == 127) {
    console.log("Couldn't find do_thing script, trying do_other_thing");
  else {
    fail('Fatal error: ' + msg, code);

Using the evented Exec object gives you a lot more flexibility in running shell commmands. But if you need something more sophisticated, Procstreams ( might be a good option.

Logging and output

Using the -q/--quiet flag at the command-line will stop Jake from sending its normal output to standard output. Note that this only applies to built-in output from Jake; anything you output normally from your tasks will still be displayed.

If you want to take advantage of the -q/--quiet flag in your own programs, you can use jake.logger.log and jake.logger.error for displaying output. These two commands will respect the flag, and suppress output correctly when the quiet-flag is on.

You can check the current value of this flag in your own tasks by using jake.program.opts.quiet. If you want the output of a jake.exec shell-command to respect the quiet-flag, set your printStdout and printStderr options to false if the quiet-option is on:

task('echo', {async: true}, function () {
  jake.exec(['echo "hello"'], function () {
  }, {printStdout: !jake.program.opts.quiet});


Jake's FileList takes a list of glob-patterns and file-names, and lazy-creates a list of files to include. Instead of immediately searching the filesystem to find the files, a FileList holds the pattern until it is actually used.

When any of the normal JavaScript Array methods (or the toArray method) are called on the FileList, the pending patterns are resolved into an actual list of file-names. FileList uses the minimatch module.

To build the list of files, use FileList's include and exclude methods:

var list = new jake.FileList();
list.include(['bar/*.txt', '']);
list.include('Makefile', 'package.json');

The include method can be called either with an array of items, or multiple single parameters. Items can be either glob-patterns, or individual file-names.

The exclude method will prevent files from being included in the list. These files must resolve to actual files on the filesystem. It can be called either with an array of items, or mutliple single parameters. Items can be glob-patterns, individual file-names, string-representations of regular-expressions, or regular-expression literals.


When you create a PackageTask, it programmatically creates a set of tasks for packaging up your project for distribution. Here's an example:

packageTask('fonebone', 'v0.1.2112', function () {
  var fileList = [
  , ''
  , 'package.json'
  , 'lib/*'
  , 'bin/*'
  , 'tests/*'
  this.needTarGz = true;
  this.needTarBz2 = true;

This will automatically create a 'package' task that will assemble the specified files in 'pkg/fonebone-v0.1.2112,' and compress them according to the specified options. After running jake package, you'll have the following in pkg/:


PackageTask also creates a 'clobber' task that removes the pkg/ directory.

The PackageTask API docs include a lot more information, including different archiving options.


When you create a TestTask, it programmatically creates a simple task for running tests for your project. The first argument of the constructor is the project-name (used in the description of the task), the second (optional) argument is a list of prerequisite tasks to run before the tests, and the final argument is a function that defines the task. It allows you to specifify what files to run as tests, and what to name the task that gets created (defaults to "test" if unset).

testTask('fonebone', ['asdf', 'qwer'], function () {
  var fileList = [
  , 'lib/adapters/**/test.js'
  this.testName = 'testMainAndAdapters';

Tests in the specified file should be in the very simple format of test-functions hung off the export. These tests are converted into Jake tasks which Jake then runs normally.

If a test needs to run asynchronously, simply define the test-function with a single argument, a callback. Jake will define this as an asynchronous task, and will wait until the callback is called in the test function to run the next test.

If you name your test 'before', it will run before any of the other tests you export. You can use it for test-setup. If you name a test 'after', it will run after all the other tests have finished. You can use it for teardown. The 'before' and 'after' will only run once per test module -- not before and after each test.

If you name your test 'beforeEach', it will run before each test. You can also name a test 'afterEach' for a test that runs after each test.

Here's an example test-file:

var assert = require('assert')
  , tests;

tests = {
  'before': function () {
    // Do some setup here
, 'after': function () {
    // Do some teardown here
, 'beforeEach': function () {
    // Something to do before every test
, 'afterEach': function () {
    // Something to do after every test
, 'sync test': function () {
    // Assert something
, 'async test': function (next) {
    // Assert something else
    // Won't go next until this is called
, 'another sync test': function () {
    // Assert something else

module.exports = tests;

Jake's tests are also a good example of use of a TestTask.


When you create a WatchTask, it will watch a directory of files for changes, and run a task or set of tasks anytime there's a change.

// Assumes there's an 'assets' task
watchTask(['assets'], function () {

Run jake watch to start up the WatchTask.

By default, it will watch the current directory for these files:

[ './**/*.js'
, './**/*.coffee'
, './**/*.css'
, './**/*.less'
, './**/*.scss'

It will exclude these files:

[ 'node_modules/**'
, '.git/**'

The list of watched files is in a FileList, with the normal include/exclude API.


The PublishTask builds on top of PackageTask to allow you to do a version bump of your project, package it, and publish it to NPM (or somewhere else). Define the task with your project's name, and call include/exclude on the packageFiles FileList to create the list of files you want packaged and published to NPM. You can also pass a list of prerequisite tasks to run before defining tasks for publishing.

Here's an example from Jake's Jakefile:

publishTask('foo', ['bar', 'baz'], function () {
  , 'Jakefile'
  , ''
  , 'package.json'
  , 'lib/**'
  , 'bin/**'
  , 'test/**'

The PublishTask will automatically create a publish task which performs the following steps:

  1. Bump the version number in your package.json
  2. Commit change in git, push it to GitHub
  3. Create a git tag for the version
  4. Push the tag to GitHub
  5. Package the new version of your project
  6. Publish it to NPM
  7. Clean up the package

If you want to publish to a private NPM repository, you can specify a custom publishing command:

npmPublishTask('zerb', function () {
  , 'index.js'
  , 'package.json'

  // Publishes using the gemfury cli
  // `%filename` will be replaced with the package filename
  this.publishCmd = 'fury push %filename';

CoffeeScript Jakefiles

Jake can also handle Jakefiles in CoffeeScript. Be sure to make it so Jake knows it's in CoffeeScript.

Here's an example:

util = require('util')

desc 'This is the default task.'
task 'default', (params) ->
  console.log 'Ths is the default task.'
  jake.Task['new'].invoke []

task 'new', ->
  console.log 'ello from new'
  jake.Task['foo:next'].invoke ['param']

namespace 'foo', ->
  task 'next', (param) ->
    console.log 'ello from next with param: ' + param

James Coglan's "Jake":

Confusingly, this is a Ruby tool for building JavaScript packages from source code.

280 North's Jake:

This is also a JavaScript port of Rake, which runs on the Narwhal platform.


Licensed under the Apache License, Version 2.0 (