The CloudFormation CLI (cfn) allows you to author your own resource providers and modules that can be used by CloudFormation.

Primary documentation for the CloudFormation CLI can be found at the AWS Documentation site.

This tool can be installed using pip from the Python Package Index (PyPI). It requires Python 3. For resource types, the tool requires at least one language plugin. Language plugins are not needed to create a module type. The language plugins are also available on PyPI and as such can be installed all at once:

pip install cloudformation-cli cloudformation-cli-java-plugin cloudformation-cli-go-plugin cloudformation-cli-python-plugin

To create a project in the current directory, use the init command. A wizard will guide you through the creation.

cfn init

To refresh auto-generated code, use the generate command. Usually, plugins try to integrate this command in the native build flow, so please consult a plugin's README to see if this is necessary. In a module project, this will regenerate the module schema.

cfn generate

To register a resource provider in your account, use the submit command.

cfn submit
cfn submit --dry-run #prepares schema handler package without submitting for registration
cfn submit --set-default # if successfully registered, set submitted version to be the new default version

To run the contract tests for a resource type, use the test command.

cfn test
cfn test -- -k contract_delete_update #to run a single test
cfn test --tb=long #exhaustive, informative traceback formatting
cfn test --enforce-timeout 60 #set the RL handler timeout to 60 seconds and CUD handler timeout to 120 seconds.
cfn test --enforce-timeout 60 -- -k contract_delete_update # combine args

To validate the schema, use the validate command.

This command is automatically run whenever one attempts to submit a resource or module. Any module fragments will be automatically validated via cfn-lint, however any warnings or errors detected by cfn-lint will not cause this step to fail.

cfn validate

To build an image for a resource type. This image provides a minimalistic execution environment for the resource handler that does not depend on AWS Lambda in anyway. This image can be used during cfn invoke and cfn test instead of using sam cli.

cfn build-image
cfn build-image --image-name my-handler --executable target/myjar.jar

The resulting image can be run in a container by executing the following command:

docker run IMAGE_NAME HANDLER_ENTRYPOINT PAYLOAD
docker run my-test-resource com.my.test.resource.ExecutableHandlerWrapper PAYLOAD_JSON # Example for a java based-project

For developing, it's strongly suggested to install the development dependencies inside a virtual environment. (This isn't required if you just want to use this tool.)

python3 -m venv env
source env/bin/activate
pip install -e . -r requirements.txt
pre-commit install

If you're creating a resource type, you will also need to install a language plugin, such as the Java language plugin, also via pip install. For example, assuming the plugin is checked out in the same parent directory as this repository:

pip install -e ../cloudformation-cli-java-plugin

Linting and running unit tests is done via pre-commit, and so is performed automatically on commit. The continuous integration also runs these checks. Manual options are available so you don't have to commit):

# run all hooks on all files, mirrors what the CI runs
pre-commit run --all-files
# run unit tests only. can also be used for other hooks, e.g. black, flake8, pylint-local
pre-commit run pytest-local

If you want to generate an HTML coverage report afterwards, run coverage html. The report is output to htmlcov/index.html.

New language plugins can be independently developed. As long as they declare the appropriate entry point and are installed in the same environment, they can even be completely separate codebases. For example, a plugin for Groovy might have the following entry point:

entry_points={
    "rpdk.v1.languages": ["groovy = rpdk.groovy:GroovyLanguagePlugin"],
},

Plugins must provide the same interface as LanguagePlugin (in plugin_base.py). And they may inherit from LanguagePlugin for the helper methods - but this is not necessary. As long as the class has the same methods, it will work as a plugin.

This library is licensed under the Apache 2.0 License.