Do users of Stack need to write carefully engineered prompts to get good results, or can they just describe what they want in plain language?

The goal with Stack is to eliminate the need for prompt engineering. Whereas going directly to tools like Claude or ChatGPT still requires a well-crafted prompt to get good output, Stack is designed so users can just describe a task in plain language and get a good result, because the system has been programmed (via the 'harness layer') to work like an accountant and doesn't need a perfectly engineered prompt to understand what's being asked.