The AI Guardrail Policy Conflict Resolver

 Act as an AI safety and policy engineer. Build a tool where I paste a list of rules from a chatbot’s system prompt or guardrail policy. Parse each rule and detect conflicts, overlaps, loopholes, and ambiguous wording, for example “always answer fully” against “never discuss pricing.” Generate 20 test scenarios that exercise the rule interactions, predict which rule wins under each one, and flag cases where the outcome is undefined. Show a conflict matrix between rules and assign each conflict a severity. For every conflict, propose a rewritten rule with explicit priority ordering. End with a cleaned and numbered policy that I can copy directly into my system prompt.