Super cool! Please let me know how it goes. Since agents are so good at writing code, we think letting the agent rewrite/test the code on failure is better than

Super cool! Please let me know how it goes. Since agents are so good at writing code, we think letting the agent rewrite/test the code on failure is better than just using a prompt at runtime