Allowing more levels of subagents calls #1374
Description
Describe the feature or problem you'd like to solve
A support for calling subagent from subagent (or even further)
Proposed solution
Currently the subgent calls (via #tool:agent/runSubagent) are followed only from the entry agent. If in instructions of agent that is called as subagent - it is not followed. It'd be great to be able to allow multiple levels of subagents (it's fine if it e.g. would need an expicit declaration in the agent frontmatter)
Example prompts or workflows
E.g.: Orchestrator agent calling Research, Plan and Implement agents. Those agents asked to break down large problems in subproblems and call self recursively as subagent, to work context effective.
The above works great when custom coded, but could be simply transformed to code-less .md agents - if multiple levels of subagents would be allowed
Additional context
No response