Author rules from policy docs, then decide: a Rete engine gives the verdict, an LLM explains why.
Probing this server's capabilities…