As an MVP of the task, it could be converting the existing python codebase of the project https://github.com/qtop/qtop into a more functional style, reducing the LOC using much more list comprehensions etc. More aims can be added according to what LLM aspects are to check.
What could come to help it is, the extensive test suite which would pin behavior strongly and make ranking of results much fairer.
As an MVP of the task, it could be converting the existing python codebase of the project https://github.com/qtop/qtop into a more functional style, reducing the LOC using much more list comprehensions etc. More aims can be added according to what LLM aspects are to check.
What could come to help it is, the extensive test suite which would pin behavior strongly and make ranking of results much fairer.