In Rust, there's a controversial practice around putting unit tests in the same file as the actual code. I was put off by it at first, but I'm finding LLM autocomplete is able to be much more effective just being able to see the tests.
No clunky loop needed.
It's gotten me back into TDD.
If the LLM can't complete a task, you add a test the shows it how to do it. This is multishot incontext learning and programming by example.
As for real TDD, you start with the tests and code until they pass. I haven't used an LLM to do this in Rust yet, but in Python due its dynamic nature, it is much simpler.
You can write the tests, then have the LLM sketch the code out enough so that they pass or at least exist enough to pass a linter. Dev tools are going to feel like magic 18 months from now.
I've sometimes done the same in python. I do quite like the ergonomics.
The benefit of this approach is that you can directly test any function in the same scope without altering its visibility: it implicitly encourages you to test all functions (and design functions in a way they can be tested, as you are writing tests as you write code), not just those part of the public api contract.
Plus you can update tests, code, and comments in one go, with visibility into them at all times.