I imagine that's the case any time you have a bash alias for some tool, I think the completion script may not know that $0 is the bare tool. I didn't test it for this example though.

David Cardona that's a bug in whatever logic determined none of the paths defined in change/rules (depends on the CI tool) where touched then nothing will be deployed It's really hard to get that logic correct without some tool like Bazel that provides developer feedback when the dependency graph isn't declared correctly.

If you can use an incremental build and test tool! Testing everything on every commit is probably too resource intensive (expensive and slow). You really want to test "only what needs to be tested". This problem is called "test selection". Some ppl manually curate a mapping from directories to tests to run, then write their own logic to query git and do the triggering. That's very error prone. Bazel gives automatic test selection just based on having high cache hit rates for anything unaffected by the developers changes.

Thanks!