Dynamic naming for spawned sub-agents

What feature would you like to see?

I would like Codex to support dynamically generated or caller-provided names for spawned sub-agents.

Today, sub-agents receive generated nicknames that do not necessarily reflect the task, role, route, subsystem, or workstream they are responsible for. In multi-agent sessions, this makes it harder to quickly understand which agent is doing what, especially when several agents are working in parallel.

A useful version of this capability could allow the caller to provide a suggested display name when spawning an agent, while still letting Codex validate or adjust the final name if needed.

For example, a spawn request could include a field such as:

{
  display_name: "API route audit",
  message: "Review the API route behavior and report findings."
}

The exact API shape is less important than the capability: sub-agent names should be able to reflect the delegated task or role.

Additional information

This would improve readability and coordination in multi-agent workflows. Generated names are fine as a fallback, but when the user or caller has a clear semantic label for the workstream, Codex should preserve that label in the UI and returned spawn metadata.

Useful behavior would include:

• Use the provided/generated semantic name as the visible sub-agent label.
• Validate length, characters, and safety as needed.
• Fall back to the current generated nickname behavior when no name is provided.
• Return both the requested name and resolved visible name in spawn metadata, if they differ.