I think there should be a global timeout that spans across all jobs. Pipelines can get pretty complex and there's a plethora of reasons why a job might hang and be stuck processing. And as a customer, we end up paying the costs.

When using orb/path-filtering and orb/continue, visualizing all related jobs is important. Maybe we can consider each job is a group and have some graph like airflow TaskGroup

steps: - run: #... run tests and store XML files to a subdirectory, for example, test-results #... - store_test_results: path: test-results

A feature to add top up credits when ever it's required.