A critical LangChain core bug can turn a single LLM prompt into a secret-exfiltration or object-instantiation exploit. If you run LangChain in production, this is urgent.