Developing a Pants Plugin

This page documents how to develop a Pants plugin, a set of code that defines new Pants functionality. If you develop a new task or target to add to Pants (or to override an existing part of Pants), a plugin gives you a way to register your code with Pants.

Much of Pants' own functionality is organized in plugins; see them in src/python/pants/backend/*.

A plugin registers its functionality with Pants by defining some functions in a file in its top directory. For example, Pants' jvm code registers in src/python/pants/backend/jvm/ Pants' backend-loader code assumes your plugin has a file there.

A "Hello World" plugin

This "Hello World" plugin example shows how to register a plugin with Pants. It defines a new hello-world goal with two tasks. Here's how to create it:

  • If you don't have an existing Pants project to work with, create one. Locate its config file, typically pants.toml in the repo root.

  • Create a directory for your plugins. In this example we will use the plugins/ directory in the repo root, but there is no convention, and you can put them wherever you like.

  • In the plugins/ directory, create following filesystem structure:

  • files can be empty - you're just saying to Python that you created modules.

  • In place the following content:

    from pants.task.task import Task
    class HelloTask(Task):
        def execute(self):
    class WorldTask(Task):
        def execute(self):

Task is a simple base class for your tasks - you must implement the execute method.

  • In place the following content:
    from pants.goal.goal import Goal
    from pants.goal.task_registrar import TaskRegistrar as task
    from hello.tasks.your_tasks import HelloTask, WorldTask
    def register_goals():
        Goal.register(name="hello-world", description="Say hello to your world")
        task(name='hello', action=HelloTask).install('hello-world')
        task(name='world', action=WorldTask).install('hello-world')

This creates a new goal named hello-world, and registers the two tasks to it.

  • In pants.toml place the following content:
    pants_version = "1.26.0"
    pythonpath = ["%(buildroot)s/plugins"]
    backend_packages = ["hello"]

backend_packages defines which plugins you want to use in your project.

If you want to use custom plugins directly from source when building in the same repo, you need to put them on the pythonpath so Pants can find them.

  • You are ready to use your plugin! First try to find your goal by typing ./pants goals:
    hello-world: Say hello to your world

Now you can use your plugin by typing ./pants hello-world:

    Executing tasks in goals: hello-world
    XX:XX:XX 00:00   [hello-world]
    XX:XX:XX 00:00     [hello]Hello

    XX:XX:XX 00:00     [world]world!

    XX:XX:XX 00:00   [complete]

Note that to consume the custom plugin as a published artifact (say on PyPI), instead of directly from the repo, then instead of backend_packages and pythonpath you would set plugins:

    pants_version = "1.26.0"
    plugins = ["myorg.hello==1.7.6"]

Similarly, if your custom plugin is consumed directly from the repo, but has dependencies on published artifacts, then you list those in plugins:

    pants_version = "1.26.0"
    pythonpath = ["%(buildroot)s/plugins"]
    backend_packages = ["hello"]
    plugins = ["some.dependency==4.5.11"]

See below for more details.

Simple Configuration

If you want to extend Pants without adding any 3rd-party libraries that aren't already referenced by Pants, you can use the following technique using sources stored directly in your repo. All you need to do is name the package where the plugin is defined, and the pythonpath entry to load it from.

In the example below, the stock JvmBinary target is subclassed so that a custom task (not shown) can consume it specifically but disregard regular JvmBinary instances (using isinstance()).

  • Define a home for your plugins. In this example we'll use a top-level plugins/ directory but this is not special.

  • Create an empty plugins/hadoop/targets/ to define a python package structure for your custom hadoop target type plugin. Also create empty files in each directory up to but not including the root directory of your python package layout; in this case, just plugins/hadoop/

  • Create a python module for your custom target type in plugins/hadoop/targets/

    # plugins/hadoop/targets/
    from pants.backend.jvm.targets.jvm_binary import JvmBinary
    class HadoopBinary(JvmBinary):
  • Create plugins/hadoop/ to register the elements exposed by your plugin to BUILD file authors:

    # plugins/hadoop/
    from pants.build_graph.build_file_aliases import BuildFileAliases
    from hadoop.targets.hadoop_binary import HadoopBinary
    def build_file_aliases():
      return BuildFileAliases(
          # NB: This allows a HadoopBinary target to be created in a BUILD file using the
          # hadoop_binary alias.
          'hadoop_binary': HadoopBinary,
  • In pants.toml, add your new plugin package to the list of backends to load when pants starts. This instructs pants to load a module named hadoop.register.

    pythonpath = ['%(buildroot)s/plugins']
    backend_packages = ['hadoop']

Note that you can also set the PYTHONPATH in your ./pants wrapper script, instead of in pants.toml, if you have other reasons to do so. Either way, pants will look for a file for each backend package you list by prefixing that package with the python path; ie roughly pythonpath + backend package + is tried for each pythonpath prefix until a is found, in this example plugins/hadoop/

Examples from twitter/commons

For an example of a code repo with plugins to add features to Pants when building in that repo, take a look at twitter/commons, especially its pants-plugins directory.

The repo's pants.ini file (the format used before pants.toml) has a backend_packages entry listing the plugin packages (packages with files):

pythonpath: [
backend_packages: [

The ...jvm/extras/ file registers a checkstyle goal. To find the code for this task, come back to the pantsbuild/pants repo: Pants defines the Checkstyle task class but doesn't register it. But other Pants workspaces can register it, as twitter/commons illustrates.

Generated by publish_docs from dist/markdown/html/src/docs/howto_plugin.html 2022-12-03T01:08:59.530128