I want to see the where does the agent want to run a terminal command.
Why?
-
If it wants to do something destructive, like
rm file
, it is essential to know WHERE -
Many times the conversation ends up in directory related trial and error, wasting tokens and time.
Example:
I want the agent to run (and fix) tests. It always ends up with the agent trying to run the command in the wrong directory, then it attempts 3-5x to fix in various ways before it finally figures it.
I know where to run it, but I don’t see anywhere what directory is the agents terminal in. I also tried to tell it base path via .cursor/rules, but it seems it is being ignored.