Multi Agent versus Coding Agent
I have been looking into the Coding Agent Skill as well as Sub-Agents as a way to delegate coding tasks.
I can use tmux subagents together with the coding agent scale to launch a review or a feature implementation, but I often have issues with getting status and blocking the main chat.
My main objective is to have Codex take care of the development because it's a much better coder than Claude.
Now with the support for multi agent, I feel like the perfect set up would be to have an additional agent configured to use Codex a developer agent. Then my main Claude agent can simply delegate any coding task as a sub-agent to the coding agent. It would automatically use Codex and the sub-agent is a native feature so it would know to get status updates and act when the job is finished.
What do you think about my thoughts here? Do you have any input suggestions? Am I on a right path? Or is there anything else I should consider?
I can use tmux subagents together with the coding agent scale to launch a review or a feature implementation, but I often have issues with getting status and blocking the main chat.
My main objective is to have Codex take care of the development because it's a much better coder than Claude.
Now with the support for multi agent, I feel like the perfect set up would be to have an additional agent configured to use Codex a developer agent. Then my main Claude agent can simply delegate any coding task as a sub-agent to the coding agent. It would automatically use Codex and the sub-agent is a native feature so it would know to get status updates and act when the job is finished.
What do you think about my thoughts here? Do you have any input suggestions? Am I on a right path? Or is there anything else I should consider?
