Indirect Prompt Injection in LangChain ReAct Agent Allows Remote Code Execution via Malicious Web Content | AI Breaking Wire