Hatch configuration¶

Configuration for Hatch itself is stored in a config.toml file located by default in one of the following platform-specific directories.

You can select a custom path to the file using the --config root option or by setting the HATCH_CONFIG environment variable.

The file can be managed by the config command group.

Mode¶

The mode key controls how Hatch selects the project to work on.

Local¶

mode = "local"

By default, Hatch will look for a pyproject.toml or setup.py file in the current working directory and any parent directories. The directory storing the first found file will be considered the project root.

Project¶

mode = "project"
project = "proj1"

[projects]
proj1 = "/path/to/project1"
proj2 = {"location": "/path/to/project2"}

[dirs]
project = ["/path/to/monorepo1", "/path/to/monorepo2"]

In this mode, Hatch will only work on the selected project. The project is located using multiple heuristics:

1. If the project is defined in the projects table then it must be a string, or an inline table with a location key, that is the full path to the project.
2. If the project matches a subdirectory in any of the directories listed in dirs.project, then that will be used as the project root.

An error will occur if the project cannot be found.

You can use the config set command to change the project you are working on:

$ hatch config set project proj2
New setting:
project = "proj2"

The project can be selected on a per-command basis with the -p/--project (environment variable HATCH_PROJECT) root option.

Aware¶

mode = "aware"

This is essentially the local mode with a fallback to the project mode.

Shell¶

You can control the shell used to enter environments with the shell key.

If defined as a string, it must be the name of one of the supported shells and be available along your PATH.

shell = "fish"

If the executable name of your shell differs from the supported name, you can define the shell as a table with name and path keys.

[shell]
name = "powershell"
path = "pwsh"

On non-Windows systems, you can change the default arguments used to spawn most shells with the args key. The default for such supported shells is usually ["-i"].

[shell]
name = "bash"
args = ["--login"]

Supported¶

Shell	Name	Arguments	macOS	Windows	Unix
Bash	bash	["-i"]
Command Prompt	cmd
C shell	csh	["-i"]
fish	fish	["-i"]
Nushell	nu	[]
PowerShell	powershell
tcsh	tcsh	["-i"]
xonsh	xonsh
Z shell	zsh	["-i"]

Default¶

On Windows systems Hatch will use the SHELL environment variable, if present, followed by the COMSPEC environment variable, defaulting to cmd. On all other platforms only the SHELL environment variable will be used, defaulting to bash.

Directories¶

Data¶

[dirs]
data = "..."

This is the directory that is used to persist data. By default it is set to one of the following platform-specific directories.

You can select a custom path to the directory using the --data-dir root option or by setting the HATCH_DATA_DIR environment variable.

Cache¶

[dirs]
cache = "..."

This is the directory that is used to cache data. By default it is set to one of the following platform-specific directories.

You can select a custom path to the directory using the --cache-dir root option or by setting the HATCH_CACHE_DIR environment variable.

Environments¶

[dirs.env]
<ENV_TYPE> = "..."

This determines where to store environments, with every key being the type of environment and the value being the desired storage location.

For example, if you wanted to store virtual environments in a .virtualenvs directory within your home directory, you could specify the following:

[dirs.env]
virtual = "~/.virtualenvs"

Any environment variables are also expanded. If the path is not absolute, then it will be relative to the project root.

Any type of environment that is not explicitly defined will default to <DATA_DIR>/env/<ENV_TYPE>.

Python installations¶

[dirs]
python = "..."

This determines where to install specific versions of Python, with the full path being <VALUE>/pythons.

The following values have special meanings.

Terminal¶

You can configure how all output is displayed using the terminal.styles table. These settings are also applied to all plugins.

[terminal.styles]
error = "..."
...

Cross-platform terminal capabilities are provided by Rich.

Output levels¶

The levels of output are as follows. Note that the verbosity indicates the minimum level at which the output is displayed.

See the documentation and color reference for guidance on valid values.

Spinner¶

You can select the sequence used for waiting animations with the spinner option.

[terminal.styles]
spinner = "..."