Pixi manages project environments, dependencies, and command-line tasks. Project settings are stored in pixi.toml, so the environment can be recreated across machines.
Initialize a project¶
Create a project in a new directory with pixi init:
pixi init my-project
cd my-projector in the folder where you are already in:
pixi initPixi creates a pixi.toml configuration file. Run project commands with pixi run so they use the project’s environment.
Add dependencies¶
Use pixi add to add packages to the project:
pixi add python
pixi add numpyPixi records dependencies in the project configuration and updates its lockfile to keep the resolved environment reproducible. Project dependencies are available to commands run through Pixi.
Define and run tasks¶
Tasks are named commands configured in pixi.toml. For example:
[tasks]
hello = "python -c 'print(\"Hello from Pixi!\")'"Run the task from the project directory:
pixi run helloTasks are useful for repeatable commands such as tests, formatting, or starting an application. For this project we created already:
pixi run editor which launches Jupyter Lab
pixi run preview which runs a Jupyter Book server and shows your project using a local host.
pixi run pdf which makes a pdf version of your project.
Install tools globally¶
Use pixi global install for a command-line tool you want to use across projects, rather than as a dependency of one project:
pixi global install python==3.14This will ensure that python is installed on your machine and can be reached throughout. We recommend to install globally:
pixi global install python==3.14 mystmd numpy matplotlib scipy jupytext jupyterlab
Note that we want to use jupyterlab_myst but this is a pypi packages. Pypi packages cannot be installed globally.
If we want to see what is installeed globally, run pixi global list. If we want to remove a package, run pixi global remove <package>. Use pixi add for project dependencies and global installs for standalone tools.