Basically rather than relying on a STEPS_DIR environment variable which has been pre-seeded, the agent should be fetching the steps that it needs to execute the pipeline.
For the first implementation this could be stupid and fetch them every time.
Steps should also be versionable on the agent side
See #35
Basically rather than relying on a
STEPS_DIR
environment variable which has been pre-seeded, the agent should be fetching the steps that it needs to execute the pipeline.For the first implementation this could be stupid and fetch them every time.
Steps should also be versionable on the agent side