The reliable countermeasure is "stop using LLMs, and build reliable software instead".

	▲	veganmosfet 3 days ago \| parent [-]
		Is the CaMel paper's idea implemented in some available agents?