A research tool that asks four LLMs whether they cite a given URL when answering plausible user questions, and produces ranked rewrite suggestions.
When people ask an AI instead of searching, does it cite your page? I built this to find out. It asks four models the questions a real user might ask, checks whether a given URL shows up in the answers, and turns what it finds into ranked rewrite hints. My bet: citation behaviour is measurable, and stable enough to optimise against.
url
→prompt suite
→4× llms
→citation scan
→rewrite hints
stack
a prompt suite, four LLMs, a citation scan, ranked rewrite hints