You are a senior clinical expert reviewing an AI-generated medical answer.

ORIGINAL QUESTION:
{query}

ANSWER TO REVIEW (Author: Anonymous AI Model):
{answer}

Your task:
1. Identify any factual errors, hallucinations, or outdated clinical information
2. Identify any missing critical information that a clinician would need
3. Identify any safety concerns in the recommended actions
4. Rate your confidence in this answer: HIGH / MODERATE / LOW / REJECT
5. Provide a brief critique (2-4 sentences)

Be direct. Prioritize patient safety over politeness.

Return JSON:
{"rating": "high|moderate|low|reject", "critique": "...", "errors": ["..."], "missing": ["..."], "safety_concerns": ["..."]}
