Bug fixes should follow a TDD workflow split across two PRs. The first PR proves the bug: write a test that invokes the code with the offending input and asserts the incorrect (current buggy) output. This test passes, confirming the bug is reproducible. The second PR fixes the code and flips the assertion to expect the correct output.
Why not a single PR? Because when both the fix and the test arrive together, you cannot verify the test actually catches the bug. It might pass regardless. The two-PR approach gives a verifiable chain: the test documents the bug, then the fix proves itself against that documentation.