It's great that when a command fails I am dropped into a shell. However if the failing command relied on environment variables, those variables aren't set and so I can't reproduce the same failure. There are two cases when this happens. 1) During an override such as an override-pull, where I expect SNAPCRAFT_* to be set (from https://snapcraft.io/docs/t/environment-variables-that-snapcraft-exposes/7569) 2) Inside a command that snapcraft itself ran. For example, if "snapcraftctl pull" runs "pip install" which fails, I expect to be able to run the command that failed (from the debug output) in the shell and see the same failure. This doesn't happen because the Python-specific environment variables aren't propagated.