So I guess I am sorta just documenting my process here:
- Learned that Claude Code just randomly decides outputs are fine if you don’t score and validate that externally (All LLMs do that sorta)
- Made sure to define strict rules and told the ai about them
- got really mad for 2 hours
- produced a website (
- Learned that Claude Code just randomly decides outputs are fine if you don’t score and validate that externally (All LLMs do that sorta)
- Made sure to define strict rules and told the ai about them
- got really mad for 2 hours
- produced a website (