[sublimefire]

It looks like the message here is “make sure to use $100k worth of Claude when doing any analysis or evaluation” and the given examples show that prior effort could have been improved or made faster. But to me 100k is an opportunity cost, and there is a possibility that these results are not reproducible, so spending it on some researcher or a grad student would buy you more in a long term. If it was 1k then sure it is worth throwing at a large problem space to find things, like using fuzzing.