Getting started

This document presents a whirlwind tour of Invoke’s feature set. Please see the links throughout for detailed conceptual & API docs. For installation help, see the project’s installation page.

Defining and running task functions

The core use case for Invoke is setting up a collection of task functions and executing them. This is pretty easy – all you need is to make a file called importing the task decorator and decorating one or more functions. You will also need to add an arbitrarily-named context argument (convention is to use c, ctx or context) as the first positional arg. Don’t worry about using this context parameter yet.

Let’s start with a dummy Sphinx docs building task:

from invoke import task

def build(ctx):

You can then execute that new task by telling Invoke’s command line runner, invoke, that you want it to run:

$ invoke build

The function body can be any Python you want – anything at all.

Parameterizing tasks

Functions can have arguments, and thus so can tasks. By default, your task functions’ args/kwargs are mapped automatically to both long and short CLI flags, as per the CLI docs. For example, if we add a clean argument and give it a boolean default, it will show up as a set of toggle flags, --clean and -c:

def build(ctx, clean=False):
    if clean:


$ invoke build -c
$ invoke build --clean

Naturally, other default argument values will allow giving string or integer values. Arguments with no default values are assumed to take strings, and can also be given as positional arguments. Take this incredibly contrived snippet for example:

def hi(ctx, name):
    print("Hi %s!" % name)

It can be invoked in the following ways, all resulting in “Hi Jeff!”:

$ invoke hi Jeff
$ invoke hi --name Jeff
$ invoke hi --name=Jeff
$ invoke hi -n Jeff
$ invoke hi -nJeff

Adding help for parameters

Describing the meaning of an argument can be done through the task’s help argument (in addition to optionally giving task-level help via the docstring):

@task(help={'name': "Name of the person to say hi to."})
def hi(ctx, name):
    """Say hi to someone."""
    print("Hi %s!" % name)

This description will show up when invoking --help:

$ invoke --help hi
Usage: inv[oke] [--core-opts] hi [--options] [other tasks here ...]

  Say hi to someone.

  -n STRING, --name=STRING   Name of the person to say hi to.

More details on how all this works can be found in the CLI concepts (for the command-line & parsing side of things) and the task API documentation (for the declaration side).

Listing tasks

You’ll sometimes want to see what tasks are available in a given tasks.pyinvoke can be told to list them instead of executing something:

$ invoke --list
Available tasks:


This will also print the first line of each task’s docstring, if it has one. To see what else is available besides --list, say invoke --help.

Running shell commands

Many use cases for Invoke involve running local shell commands, similar to programs like Make or Rake. This is done via the run function:

from invoke import task, run

def build(ctx):"sphinx-build docs docs/_build")

You’ll see the command’s output in your terminal as it runs:

$ invoke build
Running Sphinx v1.1.3
loading pickled environment... done
build succeeded, 2 warnings.

run returns a useful Result object providing access to the captured output, exit code, and so forth; it also allows you to activate a PTY, hide output (so it is captured only), and more. See its API docs for details.

Declaring pre-tasks

Tasks may be configured in a number of ways via the task decorator. One of these is to select one or more other tasks you wish to always run prior to execution of your task, indicated by name.

Let’s expand our docs builder with a new cleanup task that runs before every build (but which, of course, can still be executed on its own):

from invoke import task

def clean(ctx):"rm -rf docs/_build")

def build(ctx):"sphinx-build docs docs/_build")

Now when you invoke build, it will automatically run clean first.


If you’re not a fan of the implicit “positional arguments are pre-run task names” API, you can simply give the pre kwarg: @task(pre=[clean]).

Details can be found in the execution conceptual docs.

Creating namespaces

Right now, our is implicitly for documentation only, but maybe our project needs other non-doc things, like packaging/deploying, testing, etc. At that point, a single flat namespace isn’t enough, so Invoke lets you easily build a nested namespace. Here’s a quick example.

Let’s first rename our to be; no other changes are needed there. Then we create a new, and for the sake of brevity populate it with a new, truly top level task called deploy.

Finally, we can use a new API member, the Collection class, to bind this new task and the docs module into a single explicit namespace. When Invoke loads your task module, if a Collection object bound as ns or namespace exists it will get used for the root namespace:

from invoke import Collection, task
import docs

def deploy(ctx):"python sdist register upload")

namespace = Collection(docs, deploy)

The result:

$ invoke --list
Available tasks:


For a more detailed breakdown of how namespacing works, please see the docs.