+1 for tig, which is an awesome tool. I tend towards it mostly for walking through commit history, but it also works very nicely for staging files selectively (and is very good for selecting individual hunks in a larger changeset).
Command line+SourceTree, depending on what task I'm trying to accomplish. Staging, diffs, and commit messages are much more easily done in SourceTree, and most other operations from the command line.
Command line unless cherry picking or some other godawfully complex rare mental backflip is required, in which case SourceTree (for which I was a paid customer before it was bought by Atlassian, and responsible for the feature request leading to some of the generic hook functionality which I feel makes it very powerful for certain complex use cases)
I use hg-git. Because "hg addremove *; hg commit; hg push;" is so much simpler than the one letter alternatives in git.
Command line most of the time, with occasional use of the Git Extensions GUI client for diffs, and for committing specific hunks at a time (https://github.com/gitextensions/gitextensions)
Command line, the only way to know what you are doing
I generally use the command line, with mostly two exceptions: diffs and log histories in the IDE.