firstly I have no idea why your reply was hidden. below is it
Thanks — I suspect it involves the interaction between the orchestrator and the subagents, rather than just subagent creation. Subagents can be created, but the overall workflow still gets stuck cycling through coordination instead of making meaningful progress.
The affected workflow involves multi-step implementation, testing and review. I haven’t isolated whether task length, task type or the number of handoffs triggers it, so I can’t yet say it is exclusive to longer projects.
What needs investigating is whether the subagents themselves stall, Astra fails to recognise or act on their results, or the coordination between them breaks down. I don’t have enough runtime evidence to distinguish those possibilities yet.
So yes, an orchestration-state issue is plausible, but I wouldn’t rule out the subagent execution/handoff side. Successfully creating workers doesn’t establish that the rest of the delegated workflow is functioning correctly.