GitHub GH-200: Author and Maintain Actions Practice Questions

'act' (github.com/nektos/act) is the purpose-built open-source tool for running GitHub Actions workflows locally. It uses Docker to emulate the runner environment, reads your workflow YAML, and executes jobs on your machine. 'nektos/local-runner' is a misconception — 'act' is the actual project under the nektos organization, but it is named 'act', not 'local-runner'. 'github-runner-local' does not exist as a tool. docker-compose orchestrates multi-container Docker applications and has no awareness of GitHub Actions workflow syntax.

core.getInput('target-branch') is the correct and idiomatic way to retrieve a named input in a JavaScript action using the @actions/core toolkit. It reads the INPUT_TARGET-BRANCH environment variable under the hood and trims whitespace. process.env.TARGET_BRANCH would look for a plain environment variable named TARGET_BRANCH, not the INPUT_ prefixed one the runner sets; the exact var name would be INPUT_TARGET-BRANCH, making this unreliable without careful casing. github.context does not expose action inputs — it provides event/repository context. core.getSecret() is not a real @actions/core API method.

core.setOutput('result', value) is the correct API to publish a named output from an action so that subsequent steps can reference it via ${{ steps.<id>.outputs.result }}. core.exportVariable() sets an environment variable that is available to later steps, not an output. core.setSecret() masks a value in logs but does not create an output. core.info() merely prints to the log and does not create any accessible output value.

Within a composite action's steps, steps do not automatically halt on failure the same way a job might — especially if a prior step uses continue-on-error. Adding if: success() to each subsequent step ensures they are skipped when any prior step has failed. Composite action steps do not universally stop on first failure by default, particularly when steps use different shells or continue-on-error is involved — so claiming it is automatic is incorrect. There is no composite-action-level continue-on-error flag in action.yml. set -e in bash applies only within that specific run step's shell process and does not affect other steps in the composite action.

The valid 'using' values for the 'runs' key are JavaScript runtime versions (e.g., node20, node16), 'docker', and 'composite'. 'node20' specifies that the action runs as a JavaScript action on the Node.js 20 runtime. 'composite' specifies a composite action made up of multiple steps. 'python3' is not a supported runtime for GitHub Actions' 'using' field — Python scripts must be invoked via a Docker or composite action. 'bash' is not a valid 'using' value; shell commands are used inside composite action steps, not as a top-level runtime. 'runner' is not a valid value.