I created a skill to do that: The AI runs a typical audit process on evidence you provide. I tested this on probo as open-source evidence hub (not affiliated with them, just happened to use it, also works with other evidence sources but didn't test it)