4 ms·
Are these 'safer' than using langchain-based agents that directly execute (arbitrary!) Python code? That was always my main issue with langchain
by a_bonobo 3y ago
Are these 'safer' than using langchain-based agents that directly execute (arbitrary!) Python code? That was always my main issue with langchain