Skip to main content

In its highest aspirations, pytask tries to be pytest as a build system.

Project description

PyPI PyPI - Python Version PyPI - License status

In its highest aspirations, pytask tries to be pytest as a build system. It’s main purpose is to facilitate reproducible research by automating workflows in research projects. Its features include:

  • Automatic discovery of tasks.
  • Lazy evaluation. If a task, its dependencies, and its products have not changed, do not execute it.
  • Debug mode. Jump into the debugger if a task fails, get feedback quickly, and be more productive.
  • Select tasks via expressions. Run only a subset of tasks with expressions and marker expressions known from pytest.
  • Easily extensible with plugins. pytask’s architecture is based on pluggy, a plugin management framework, so that you can adjust pytask to your needs. Plugins are, for example, available for parallelization, LaTeX, R, and Stata. Read here how you can use plugins.


pytask is available on PyPI for Python >= 3.6.1 and on for Python >= 3.7. Install the package with

$ pip install pytask

# or

$ conda install -c conda-forge pytask

Color support is automatically available on non-Windows platforms. On Windows, please, use Windows Terminal which can be, for example, installed via the Microsoft Store.


A task is a function which is detected if the module and the function name are prefixed with task_. Here is an example.

# Content of

import pytask

def task_hello_earth(produces):
    produces.write_text("Hello, earth!")

Here are some details:

  • Dependencies and products of a task are tracked via markers. For dependencies use @pytask.mark.depends_on and for products use @pytask.mark.produces. Use strings and pathlib.Path to specify the location. Pass multiple dependencies or products as a list or a dictionary for positional or key-based access.
  • With produces (and depends_on) as function arguments, you get access to the dependencies and products inside the function via pathlib.Path objects. Here, produces holds the path to "hello_earth.txt".

To execute the task, type the following command on the command-line

$ pytask
========================= Start pytask session =========================
Platform: linux -- Python 3.x.y, pytask 0.x.y, pluggy 0.x.y
Root: xxx
Collected 1 task(s).

====================== 1 succeeded in 1 second(s) ======================


The demo walks you through the following steps.

  1. Write an executable script which produces a text file like you would normally do without pytask.
  2. Rewrite the script to a pytask task.
  3. Execute the task.
  4. Add a task which produces a second text file and another task which merges both text files.
  5. Execute all three tasks.


The documentation can be found under with tutorials and guides for best practices.


Consult the release notes to find out about what is new.


pytask is distributed under the terms of the MIT license.


The license also includes a copyright and permission notice from pytest since some modules, classes, and functions are copied from pytest. Not to mention how pytest has inspired the development of pytask in general. Without the amazing work of Holger Krekel and pytest’s many contributors, this project would not have been possible. Thank you!


If you rely on pytask to manage your research project, please cite it with the following key to help others to discover the tool.

    Title  = {A Python tool for managing scientific workflows.},
    Author = {Tobias Raabe},
    Year   = {2020},
    Url    = {}

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for pytask, version 0.1.0
Filename, size File type Python version Upload date Hashes
Filename, size pytask-0.1.0-py3-none-any.whl (75.7 kB) File type Wheel Python version py3 Upload date Hashes View
Filename, size pytask-0.1.0.tar.gz (66.5 kB) File type Source Python version None Upload date Hashes View

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring DigiCert DigiCert EV certificate Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page