Silly guy has been diving into openclaw crayfish for a few days now. In addition to the joy of smooth progress, we also encountered many strange problems. Even many problems are not encountered by local IDE programming. For example, today's self media business debugging lobster. The message was sent to the wrong Telegram conversation, and many reasons were found Later on, the wealth line sessions only defined the chat ID without specifying the agent ID For example, combining multiple models, optimizing strong models, and reserving weak models But once the strong model GPT5.4 responds, it will be slightly slower The weak model is on top It seems that the system does not have a mechanism for strong models to repeatedly retry For example, to analyze certain sources of news information Will encounter errors in various fields and summaries Despite various minor issues But Silly Brother has always believed that none of these can stop him Training crayfish is like adjusting the chassis of a car; Requires skills, techniques, and accumulated experience Sharpening your ax will not delay your job of cutting wood; It will be a sharp knife