who the f*ck let a spider fix prompts... why am i still doing this manually someone ran a 181-word prompt through it looked clean had been used for weeks first score: 37/100 the spider read it without assuming it knew what anything meant → tagged every word as a claim,