When creating unit tests, do you rely on pre-prepared test scenarios or develop them independently during the process?

When creating unit tests, I usually combine both approaches. If there are pre-prepared test scenarios, I use them as a basis to cover the main cases and requirements. However, during the development process, additional boundary cases or unexpected situations often emerge, which require independent development of new tests.

For example, when testing a data processing function, I might start with basic scenarios from the specification, then add tests for error handling or non-standard input data that were not explicitly described in the documentation.