Geometric tests are here https://github.com/psi4/psi4/blob/master/tests/pytests/test_geometric.py , but they wouldn't have caught the problem you identified. If you feel like adding more, go for it, but I don't think this is a problem we're likely to relapse on.

If you wanted to run those, from a compiled psi4 in <objdir>, execute the results of stage/bin/psi4 --psiapi to set up paths, then pytest -v ../tests/pytests/test_geometric.py

The only way this bug can re-appear is if geometric will change the way they process the atomic labels (which is what happened I suppose). Let me know if a test with Cl atom has to be added.

The only way this bug can re-appear is if geometric will change the way they process the atomic labels (which is what happened I suppose). Let me know if a test with Cl atom has to be added.

No need, imo. It's psi sending atoms weirdly (for legacy case-code-is-longwinded-in-old-c++ reasons), not geometric expecting atoms weirdly.

The CI fails in the _________________________ test_qcf_cbs_mbe[snowflake] __________________________ section but I do not think it is related to my fix.

test_qcf_cbs_mbe[snowflake]

correct, I'm not sure right off if it's a fluke or if qcfractal has changed, but your PR is innocent.

test_qcf_cbs_mbe[snowflake]

correct, I'm not sure right off if it's a fluke or if qcfractal has changed, but your PR is innocent.

It does not look like fluke, it has been failing on my PRs for a while, eg. #2864