Yup, both Claude v5 models to me feel like they have extreme ADD or something. Using them feels like walking a dog that was never leash trained and constantly needs to be kept moving in the right direction.
I think it's optimized around beating benchmarks and running fully autonomous in pursuit of a clearly defined goal. It makes sense: fan out aggressively, chase down every lead, but go depth first because that's easier for the LLM and you're either a sub-agent with a narrowly defined task or you're a top-level orchestrator agent with a /goal loop that will catch and fix errors and omissions on the second, third, fourth pass.
I think it's optimized around beating benchmarks and running fully autonomous in pursuit of a clearly defined goal. It makes sense: fan out aggressively, chase down every lead, but go depth first because that's easier for the LLM and you're either a sub-agent with a narrowly defined task or you're a top-level orchestrator agent with a /goal loop that will catch and fix errors and omissions on the second, third, fourth pass.