The correct option is To define a specific run order so that one job waits for another to complete. This ensures that the build job can finish and publish artifacts before the test job begins so the artifacts are available when tests run.

In GitHub Actions you declare a dependency by using the needs key on the dependent job. When the test job needs the build job the runner schedules the test only after the build has completed which creates a clear execution order and enforces waiting. This is the primary mechanism for orchestrating multi job workflows where outputs and artifacts must flow from one job to another.

To set and enforce maximum concurrent runs across the entire workflow is not what job dependencies do. Limiting concurrency for a workflow or for a job is handled by the concurrency key which groups runs and cancels or queues them and it is unrelated to needs based dependencies.

To evaluate complex conditions based on previous job results and expressions is not the primary role of dependencies. You can reference the needs context in if expressions to make conditional decisions about running a job or a step yet the dependency itself simply defines which job must complete first.

To enable jobs to execute in parallel without waiting on others is the opposite of what a dependency does. Jobs without a dependency run in parallel by default whereas declaring a dependency makes the next job wait.

Full AWS Practitioner Certification Question