Right now kanna does create the sub agents through codex or claude code, but those sub agents can not be visualized like they are done through codex desktop app. Whenever a sub agent is spawned up, the relevant small box should be shown so that when the end user clicks on it, a new tab on any suitable position opens up to see the activity happening in that sub agent only. This helps in understanding what each sub agent is doing, instead of guessing from the thinking outputs visible in the main thread.
Right now kanna does create the sub agents through codex or claude code, but those sub agents can not be visualized like they are done through codex desktop app. Whenever a sub agent is spawned up, the relevant small box should be shown so that when the end user clicks on it, a new tab on any suitable position opens up to see the activity happening in that sub agent only. This helps in understanding what each sub agent is doing, instead of guessing from the thinking outputs visible in the main thread.