With the introduction of more in-depth subagent spawning, and async processing, it’s allowed me to make very interesting multi-agent setups I’ve wanted since I saw subagents were released in cursor, but realized were not possible, as subagents could not call upon other subagents, and only one could run at a time.
It’s been clear since the release of subagents that auto, and composer models are not allowed to spawn their own subagents from chat, but now with the implementation of composer 1.5 I think there’s little reason besides it being tied to auto usage, that it cannot spawn subagents.
A side-effect of this is that you can only do subagent tasks on paid API models, which is fair, BUT if you wanted to use composer 1.5 which is a great model IMO as the subagent actor it becomes a dead-end because it cannot carry out any further subagent spawns.
My suggestion is that it be made clear when you select these composer 1, composer 1.5, and auto models that you are not able to use subagents, and that they will not be able to spawn subagents if selected as the actor of a agents.md file.
Or if you guys wanted to be dope asf let me spawn subagents with comp 1.5 because its my main model now, the context management is nuts!