Installation

pipescript is published on PyPI. The most common way to install it is from inside a running IPython or Jupyter session, so that the same session can load the extension:

%pip install pipescript
%load_ext pipescript

The %load_ext pipescript line is what actually enables the new syntax – the pipe operators, placeholders, and macros – in your current session. Without it, pipescript is installed but dormant.

You can equally install from a normal shell:

$ pip install pipescript

and then run %load_ext pipescript once inside IPython/Jupyter.

Runtime dependencies are minimal: pyccolo (which does the source-to-source instrumentation – see How it works) and typing-extensions.

Supported environments

pipescript is currently only for interactive Python environments built on top of IPython – that is, IPython itself and Jupyter (Notebook, Lab, and compatible front-ends such as ipyflow). It is not a standalone language and has no effect in a plain python script run outside of IPython.

It supports Python 3.9 through 3.14. Because instrumentation is embedded at the level of source code rather than bytecode, the same syntax works across that whole range.

Trying it without installing

If you just want to experiment, a full pipescript environment runs entirely in your browser via JupyterLite – no local install:

https://smacke.github.io/pipescript/lab/index.html?path=demo.ipynb

Working from a source checkout

If you are hacking on pipescript itself, clone the repository and install it in editable mode with the relevant extras:

$ git clone https://github.com/smacke/pipescript.git
$ cd pipescript
$ pip install -e '.[test]'   # test/lint/type-check toolchain
$ pip install -e '.[docs]'   # Sphinx toolchain for building these docs

You can then build this documentation locally:

$ make -C docs html

The rendered site lands in docs/_build/html.

Next steps

With pipescript installed and loaded, head to Your first pipeline to build something that runs.