Python
Kloud Workspace ships a Python toolchain that is ready to use:
- Python 3.13 runtime.
- Both
pipand theuvpackage manager. - The
pixipackage manager for conda and PyPI environments. - Automatic virtual environment activation when you
cdinto a project directory. - Linting and formatting with
ruffand its VS Code extension. - Code intelligence from Meta's Pyrefly language server.
- An
IPythonprofile in the Kloud Workspace colors (IPython itself is not preinstalled).
Additional Packages & Tools
When a new container starts, the workspace installs the Python packages and CLI tools listed in two variables:
WS_PIP_ADDITIONAL_PACKAGES: Additional global Python packages installed during startup.WS_UV_ADDITIONAL_TOOLS: Additional global Python CLI tools installed during startup.
Both accept a space-delimited list. Use pip to install importable libraries into the user site, and uv to install global CLI tools.
docker run \
-e WS_PIP_ADDITIONAL_PACKAGES="httpx==0.27.0 rich" \
-e WS_UV_ADDITIONAL_TOOLS="poetry==1.8.0 cookiecutter" \
ghcr.io/kloudkit/workspace:v0.6.0Package Management with uv
Use uv for day-to-day dependency work. It resolves dependencies fast, installs reproducibly from a lock file, and works with requirements.txt and pyproject.toml.
Automatic venv Activation
When you cd into a directory that contains a .venv or venv folder, the auto-venv plugin activates its virtual environment, and leaving the project deactivates it. It checks only the directory you cd into: a terminal opened inside a project, or a cd straight into a subdirectory, activates nothing. It works with environments created by uv venv or python -m venv.
To activate a new environment right away, create it at the project root and run cd .:
uv venv
cd .Conda Environments with pixi since v0.6.0
pixi manages per-project conda and PyPI environments from a pixi.toml or pyproject.toml. Tools you install with pixi global install land in ~/.local/pixi/bin, which is on PATH. For settings every user shares, such as a channel mirror, add /etc/pixi/config.toml.
Linting & Formatting with ruff
Kloud Workspace bundles Ruff and its VS Code extension, for feedback as you type. The default configuration is opinionated:
- A cache in
~/.cache/ruff. - Fixes applied automatically.
- An indent width of 2 and a maximum line length of 80.
- Python 3.13 as the target version.
- A ruleset that selects and ignores specific error codes from several linting plugins.
- Formatting with
lfline endings, including code in docstrings. - Rules for import sorting and type-checking blocks.
- Per-file overrides that turn off the docstring rules in test directories.
Configuration Lookup
The default Ruff configuration lives at ~/.config/ruff/ruff.toml and suits most projects. For custom linting rules, add a ruff.toml or .ruff.toml file at your project's root: it replaces the default configuration instead of extending it.
To start from the default, copy it into the project as .ruff.toml:
ws-cli template apply ruffLanguage Server
Meta's Pyrefly is the Python language server, with strict type checking on by default. It honors the standard Pyrefly configuration.
For a lighter language server, override the editor settings with WS_EDITOR_SETTINGS_MERGE: turn off python.pyrefly.* and select the language server you prefer.
Optional Interactive Shell
For a richer REPL, install IPython:
pip install ipythonKloud Workspace already includes an IPython profile (/etc/ipython/ipython_config.py) whose prompt colors and syntax highlighting match the workspace theme.
IPython installed with pip outside a virtual environment picks up these settings. Inside a virtual environment, also install catppuccin[pygments], which the profile's colors need.