I think there is a limitation in that it cannot check the element.

For example, it would not throw an exception in the following case:

String actual = sut.giveMeBuilder(new TypeReference<List<List<SimpleObject>>>() {
			})
			.size("$", 1)
			.size("$[0]", 1)
			.set("$[0][1].str", "expected")
			.sample()
			.get(0)
			.get(0)
			.getStr();

Nevertheless, it is a useful feature, so adding a comment to describe the limitation would be better.

For example, it would not throw an exception in the following case:

here is a passing test case.

thenThrownBy(
		() -> sut.giveMeBuilder(new TypeReference<List<List<SimpleObject>>>() {
			})
			.size("$", 1)
			.size("$[0]", 1)
			.set("$[0][1].str", "expected")
			.sample()
	).isInstanceOf(IllegalArgumentException.class)
		.hasMessageContaining("No matching results for given NodeResolvers.");

As for the other test case you suggested, I felt it was already covered by the existing tests, so I didn't add it separately.
However, if you still think a more explicit test would be valuable, I'm happy to add it.

On a related note, I'm not sure about the best way to proceed with the isValidFieldPath method.
Should Iadd a comment to explain its current limitations, or would adding a test case that demonstrates the issue be sufficient?

If you think a comment is the right approach, could you please recommend what it should say?

Sorry, I've misunderstood. The above case was worked on using the existing MonkeyExpressionFactory.

I mean, in the case below, it does not throw an error even if the first element does not exist.

FixtureMonkey sut = FixtureMonkey.builder().useExpressionStrictMode().build();

		List<SimpleObject> actual = sut.giveMeBuilder(new TypeReference<List<List<SimpleObject>>>() {
			})
			.size("$", 1)
			.size("$[1]", 1)
			.sample()
			.get(0);

I think it's an edge case. It would be better to write a comment about the limitations.

Thanks for the late-night review!

I've addressed it in commit