Agno Toolkit for the Ejentum Reasoning Harness. Four cognitive harnesses (reasoning, code, anti-deception, memory) as agent-callable methods. Each call returns a task-matched cognitive operation engineered in two layers: natural-language procedure plus executable reasoning topology.