Cookiecutter: Automate project creation!


As I move closer to the world of development within my career I have been looking for more efficient ways to spend my time, along with assisting my colleagues and myself follow the programming, documenting and best practices we have set.

When we create a new project there are many repetitive tasks that take place, such as creating pyproject.toml, directory structures, documentation folders and many other tasks, these tasks are time consuming, repetitive and prone to user error.

Some context

Starting a new repository for a new project is always a chore, specially when working with large teams where others are collaborating with you. You have to follow the same standards and coding practices to ensure all developers know what is happening.

Working in large teams means that with many different projects and repositories it is very likely that none of them will follow the same base structure that is expected. To help alleviate this problem and fulfil these expectations I created project templates that anyone can follow to ensure all base projects are the same.

What is Cookiecutter

Cookiecutter is a CLI tool built in Python that creates a project from boilerplate templates (mainly available on Github). It uses the templating system Jinja2 to replace and customize folders and/or files names, as well as their content.

Although built with Python, you are not limited to templating Python projects, it can easily be implemented with other programming languages. However, to do this you will need to know or learn some Jinja and if you want to implement hooks this will need to be done in Python.

Why use cookiecutter

Well simply put, to save time building new project repositories, to avoid missing files or commit checks and probably one important step, to make life easier for new team members who will be expected to create projects.

We also use it as a way to enforce standards, providing the developer with the necessary structure to ensure the rules are followed: write documentation perform tests, follow specific syntax standards by giving them the base structure in a boilerplate code, it makes it easier for developers to follow standards.

In certain projects you may have a lot of repetitive code, such as creating Flask websites, with a cookiecutter template, you would be able to duplicate that code with ease and little time spent.

How to use Cookiecutter

Cookiecutter is super simple to use, you can either use one of the many templates that already exist online, or you can create one that suits your own needs.

You can access templates from various locations:

  • Git repository
  • Local folder
  • Zip file

If working with Git repositories, you can even start a template from any branch!

To try out cookie cutter, first it needs to be installed:

pip install -U cookiecutter

Once installed run the following command:

cookiecutter gh:totaldebug/python-package-template

This repository follows the standards that I use for my Python Package repositories. When you execute it, you will be prompted for values, for example:

full_name [marksie1988]: Steve Marks
email []:
github_username [totaldebug]:
version [0.1.0]:
use_pytest [n]: y

For a default value, you just press return, or to amend the value you simply type it and hit return. This prompt is created based on the values defined in cookiecutter.json, all cookiecutter templates have this file in the root.

Once you have answered all of the prompts, the template will be converted into a new location using the data provided during the prompts.

How Cookiecutter templates work

A basic Cookiecutter template looks like this:

 ┣ 📂hooks
 ┃ ┣ 📜
 ┃ ┗ 📜
 ┗ 📂
 ┃ ┣ 📜your-project-files-here
 ┃ ┣ 📂
 ┃ ┃ ┗ 📂3d-printer-axes-calibration
 ┗ 📜cookiecutter.json

What happens here:

template: this is the root of the repository or folder

hooks: Python scripts that execute before and after the generation of the repository. Pre-hooks are generally used to validate inputs from the prompts and the post-hooks to remove files that are not require for this specific project.

{{ cookiecutter.project_slug }}: is the directory for your project to be stored. Anything stored in this directory will be copied to the new project.

For a python package you would have another subdirectory with the package name this would usually be the {{ cookiecutter.project_slug }}/{{ cookiecutter.project_slug }} directory.

This is the minimum required file structure, you can then add as required for your projects, or copy an existing template and amend the areas that you require.


To allow flexibility with a template you add variables to cookiecutter.json this will create a prompt when executing the template for a value which will change the output to the template.

For each variable within here a default text value, boolean or list of options are required. Example:

  "full_name": "marksie1988",
  "email": "",
  "github_username": "totaldebug",
  "project_name": "Python Boilerplate",
  "project_slug": "{{ cookiecutter.project_name.lower().replace(' ', '_').replace('-', '_') }}",
  "minimal_python_version": [3.7, 3.8, 3.9],
  "use_black": "y"

Input validity can be checked with pre_gen_project hooks, the below example validates the data supplied in the project_slug value:

MODULE_REGEX = r'^[_a-zA-Z][_a-zA-Z0-9]+$'

module_name = ''

if not re.match(MODULE_REGEX, module_name):
    print('ERROR: The project slug (%s) is not a valid Python module name. Please do not use a - and use _ instead' % module_name)

As you can see from the examples, you can either create a very simple template or add Jinja / Python for more complex and error validation.

Final Thoughts

Cookiecutter has saved me a lot of time in the creation or projects, also a lot of the boring template work is taken out of starting a new project which is always a bonus.

Now all of my projects start in a good standard and should be easier to keep that way.

If you would like to check out cookiecutter you could start by checking my python-package-template

I have added things like Github actions and pre-commits to check work along with other python best practices that I hope to cover in my next article.

Hopefully some of this information was useful for you, If you have any questions about this article and share your thoughts head over to my Discord.

This post is licensed under CC BY 4.0 by the author.