Skip to main content
Version: 2.14 (deprecated)

Shell overview

Pants's support for Shellcheck, shfmt, and shUnit2.

Pants integrates with these tools to empower you to follow best practices with your Shell scripts:

  • Shellcheck: lint for common Shell mistakes.
  • shfmt: autoformat Shell code so that you can instead focus on the logic.
  • shUnit2: write light-weight unit tests for your Shell code.

Pants installs these tools deterministically and integrates them into the workflows you already use: ./pants fmt, ./pants lint, and ./pants test.

Initial setup: add shell_sources targets

Pants uses shell_source and shunit2_test targets to know which Shell files you want to operate on and to set any metadata.

To reduce boilerplate, the shell_sources target generates a shell_source target for each file in its sources field, and shunit2_tests generates a shunit2_test target for each file in its sources field.

shell_sources(name="lib", sources=["", ""])
shell_tests(name="tests", sources=[""])

# Spiritually equivalent to:
shell_source(name="deploy", source="")
shell_source(name="changelog", source="")
shell_test(name="changelog_test", source="")

# Thanks to the default `sources` values, spiritually equivalent to:

First, activate the Shell backend in your pants.toml:

backend_packages = [

Then, run ./pants tailor :: to generate BUILD files:

$ ./pants tailor ::
Created scripts/BUILD:
- Add shell_sources target scripts
Created scripts/subdir/BUILD:
- Add shell_sources target subdir

You can also manually add targets, which is necessary if you have any scripts that don't end in .sh:

shell_source(name="script_without_a_extension", source="script_without_an_extension")
Shell dependency inference

Pants will infer dependencies by looking for imports like source and . You can check that the correct dependencies are inferred by running ./pants dependencies path/to/ and ./pants dependencies --transitive path/to/

Normally, Pants will not understand dynamic sources, e.g. using variable expansion. However, Pants uses Shellcheck for parsing, so you can use Shellcheck's syntax to give a hint to Pants:


# Normally Pants couldn't infer this, but we can give a hint like this:
# shellcheck source=dir/
source "${another_script}"

Alternatively, you can explicitly add dependencies in the relevant BUILD file.


shfmt autoformatter

To activate, add this to your pants.toml:

backend_packages = [

Make sure that you also have set up shell_source/shell_sources or shunit2_test/shunit2_tests targets so that Pants knows to operate on the relevant files.

Now you can run ./pants fmt and ./pants lint:

$ ./pants lint scripts/
13:05:56.34 [WARN] Completed: lint - shfmt failed (exit code 1).
--- scripts/
+++ scripts/
@@ -9,7 +9,7 @@

set -eo pipefail

-HERE=$(cd "$(dirname "${BASH_SOURCE[0]}")" && \
+HERE=$(cd "$(dirname "${BASH_SOURCE[0]}")" &&

𐄂 shfmt failed.

Use ./pants fmt lint dir: to run on all files in the directory, and ./pants fmt lint dir:: to run on all files in the directory and subdirectories.

Pants will automatically include any relevant .editorconfig files in the run. You can also pass command line arguments with --shfmt-args='-ci -sr' or permanently set them in pants.toml:

args = ["-i 2", "-ci", "-sr"]

Temporarily disable shfmt with --shfmt-skip:

./pants --shfmt-skip fmt ::

Only run shfmt with --lint-only and --fmt-only:

./pants fmt --only=shfmt ::
Benefit of Pants: shfmt runs in parallel with Python, Java, Scala, and Go formatters

Normally, Pants runs formatters sequentially so that it can pipe the results of one formatter into the next. However, Pants will run shfmt in parallel to formatters for other languages, like Python, because shfmt does not operate on those languages.

You can see this concurrency through Pants's dynamic UI.

Shellcheck linter

To activate, add this to your pants.toml:

backend_packages = [

Make sure that you also have set up shell_source / shell_sources or shunit2_test / shunit_tests targets so that Pants knows to operate on the relevant files.

Now you can run ./pants lint:

$ ./pants lint scripts/
13:09:10.49 [WARN] Completed: lint - Shellcheck failed (exit code 1).

In scripts/ line 12:
HERE=$(cd $(dirname ${BASH_SOURCE[0]}) && pwd)
^--------------------------^ SC2046: Quote this to prevent word splitting.
^---------------^ SC2086: Double quote to prevent globbing and word splitting.

Did you mean:

𐄂 Shellcheck failed.

Use ./pants fmt lint dir: to run on all files in the directory, and ./pants fmt lint dir:: to run on all files in the directory and subdirectories.

Pants will automatically include any relevant .shellcheckrc and shellcheckrc files in the run. You can also pass command line arguments with --shellcheck-args='-x -W 3' or permanently set them in pants.toml:

args = ["--external-sources", "--wiki-link-count=3"]

Temporarily disable Shellcheck with --shellcheck-skip:

./pants --shellcheck-skip lint ::

Only run Shellcheck with --lint-only:

./pants lint --only=shellcheck ::
Benefit of Pants: Shellcheck runs in parallel with other linters

Pants will attempt to run all activated linters and formatters at the same time for improved performance, including Python, Go, Java, and Scala linters. You can see this through Pants's dynamic UI.

shUnit2 test runner

shUnit2 allows you to write lightweight unit tests for your Shell code.

To use shunit2 with Pants:

  1. Create a test file like,, or
  2. Create a shunit2_test or shunit2_tests target in the directory's BUILD file.
  3. Specify which shell to run your tests with, either by setting a shebang directly in the test file or by setting the field shell on the shunit2_test / shunit2_tests target.
    • See here for all supported shells.
#!/usr/bin/env bash

testEquality() {
assertEquals 1 1

You can then run your tests like this:

# Run all tests in the repository.
./pants test ::

# Run all the tests in the folder.
./pants test scripts:

# Run just the tests in this file.
./pants test scripts/

Pants will download the ./shunit2 script and will add source ./shunit2 with the correct relpath for you.

You can import your production code by using source. Make sure the code belongs to a shell_source or shell_sources target. Pants's dependency inference will add the relevant dependencies, which you can confirm by running ./pants dependencies scripts/ You can also manually add to the dependencies field of your shunit2_tests target.


source scripts/

testAdd() {
assertEquals $(add_one 4) 5
Running your tests with multiple shells

Pants allows you to run the same tests against multiple shells, e.g. Bash and Zsh, to ensure your code works with each shell.

To test multiple shells, use the parametrize mechanism, like this:

shell=parametrize("bash", "zsh"),

Then, use ./pants test:

# Run tests with both shells.
./pants test scripts/

# Run tests with only Zsh.
./pants test scripts/

Controlling output

By default, Pants only shows output for failed tests. You can change this by setting --test-output to one of all, failed, or never, e.g. ./pants test --output=all ::.

You can permanently set the output format in your pants.toml like this:

output = "all"

Force reruns with --force

To force your tests to run again, rather than reading from the cache, run ./pants test --force path/to/

Setting environment variables

Test runs are hermetic, meaning that they are stripped of the parent ./pants process's environment variables. This is important for reproducibility, and it also increases cache hits.

To add any arbitrary environment variable back to the process, use the option extra_env_vars in the [test] options scope. You can hardcode a value for the option, or leave off a value to "allowlist" it and read from the parent ./pants process's environment.

extra_env_vars = ["VAR1", "VAR2=hardcoded_value"]

Use [bash-setup].executable_search_paths to change the $PATH env var used during test runs. You can use the special string "<PATH>" to read the value from the parent ./pants process's environment.

executable_search_paths = ["/usr/bin", "<PATH>"]


Pants can cancel tests that take too long, which is useful to prevent tests from hanging indefinitely.

To add a timeout, set the timeout field to an integer value of seconds, like this:

shunit2_test(name="tests", source="", timeout=120)

When you set timeout on the shunit2_tests target generator, the same timeout will apply to every generated shunit2_test target. Instead, you can use the overrides field:

"": {"timeout": 20},
("", ""): {"timeout": 35},

You can also set a default value and a maximum value in pants.toml:

timeout_default = 60
timeout_maximum = 600

If a target sets its timeout higher than [test].timeout_maximum, Pants will use the value in [test].timeout_maximum.

Use the option ./pants test --no-timeouts to temporarily disable timeouts, e.g. when debugging.

Testing your packaging pipeline

You can include the result of ./pants package in your test through the runtime_package_dependencies field. Pants will run the equivalent of ./pants package beforehand and copy the built artifact into the test's chroot, allowing you to test things like that the artifact has the correct files present and that it's executable.

This allows you to test your packaging pipeline by simply running ./pants test ::, without needing custom integration test scripts.

To depend on a built package, use the runtime_package_dependencies field on the shunit2_test / shunit2_tests targets, which is a list of addresses to targets that can be built with ./pants package, such as pex_binary, python_awslambda, and archive targets. Pants will build the package before running your test, and insert the file into the test's chroot. It will use the same name it would normally use with ./pants package, except without the dist/ prefix.

For example:

python_source(name="py_src", source="")
pex_binary(name="pex", entry_point="")