Glm-5.1+ and gpt-5.5+ seem quite capable. My agents is mostly helping glm with a jj status command it would reliably try but get the quoting wrong on. It's not

Glm-5.1+ and gpt-5.5+ seem quite capable. My agents is mostly helping glm with a jj status command it would reliably try but get the quoting wrong on. It's not confused about detached head.