The issue wasn't just the incorrect sum (which I was aware of), but rather the inconsistent behavior of the assert engine. Despite making no changes to the test values or code, tests would randomly pass or fail. In some instances, even with the correct string, it would throw an error. Paradoxically, the same code with the correct sum would fail, yet running it again without modifications would suddenly pass. It was baffling.

To resolve this, I reset the code completely, reimplemented the correct solution, and it finally worked consistently.

This appears to be a platform bug. If you encounter tests arbitrarily passing or failing without code changes, remember that resetting the problem is likely your solution