It would be good to have some info about the expected resource usage of running gcpdiag (how much memory the process may consume, or network bandwidth for example). This might scale in terms of the number of resources in the project being analyzed, so a few examples may be necessary.

Also please include best practices for choosing an environment to run gcpdiag, considering

• Cloud Shell would work for most cases
• if the user needs to limit the resource usage (run within a container for example)
• whether or not to run on a production machine vs a throw-away VM.