Hi! I'm a security engineer that is looking to improve the quality of my fuzzing harness.

The goal is to look at how the corpus of the fuzzing harness can be improved.

The project is {PROJECT_NAME} and the repository of the project is {PROJECT_REPOSITORY}.

The fuzzer I'm working on is:
<code>
{HARNESS_SOURCE_CODE}
</code>

The target code of the harness is:
<code>
{TARGET_FUNCTION_SOURCE}
</code>

Could you please construct a small Python program that generate 10 initial seeds for my fuzz harness?

The corpus generator should be focused on making structured seeds that reflect the structure of the target function. It should not be based purely on the output from random.

The program you write to generate seeds should take no input and should output the seeds into the current working folder.

Wrap the program in <results> tags in the reply and do not return any other text.