Remote Code Execution

LangChain vulnerable to code injection

In LangChain through 0.0.131, the `LLMMathChain` chain allows prompt injection attacks that can execute arbitrary code via the Python `exec()` method.