stdout as context if a command is ran

[CentillionJinx]

I wonder if it is possible to have the terminal output to be part of the context of a model, when it runs a command, and it's subsequent actions are dependant on the contents of the stdout. For example, I sometimes run leak checks on my programs, and the output of valgrind is crucial for the model to determine if the code has no leaks or if there is pending issues that needs to be resolved. For most of the times, I would have to terminate the API request, right after the command is ran, and then attach the terminal output to the chat for the model to get the context of stdout.