User should be able to modify commands before they are executed by the assistant

[adevaucorbeil]

Quite often, when the LLM wants to run a command, they are not completely accurate and the LLM has to try a couple times to get it right. Often, I even have to provide the right command for it to execute.
Besides Kilocode, I also use GitHub Copilot, and in Copilot, one can modify the commands before they are executed. This is very handy and removes many frustrations for me.
Implementing this would reduce the number of tokens one need to use before the right command is executed (as long as the user knows how to improve the command that is).

Would be great to see this happening!