用户强调 Omo 和 Hermes 代理如何通过优先考虑实际意图而非基准分数来有效处理大型代码库。

do you remember the first time you worked with omo?
that the fact it even works well with large codebase, catches your intent, and produces the code that you ideally wanted to have

how come?

i used claude code and i am a person who got crazy for actual, practical use case optimizations

no benchmarks were there cuz we optimized for your use case, not for those autistic numbers that you "feel" like you are working with the best agent

i am using openclaw, hermes agent these days

and i think like i am seeing something different again

my GitHub is on fire again