In the case of executing a pipeline which will run only a subset of a pipeline, think a specific stage, the agent must have a means of executing only specific contexts and then exiting.
I think this can just be a CLI argument and then the agent can only execute steps (sequentially) which match those context IDs
I implemented this a while ago when I refactored some internal structures. The agent is now spawned only with a list of steps it can run uninterrupted.
In the case of executing a pipeline which will run only a subset of a pipeline, think a specific stage, the agent must have a means of executing only specific contexts and then exiting.
I think this can just be a CLI argument and then the agent can only execute steps (sequentially) which match those context IDs