This paper addresses the challenge of equipping automated trading agents with market impact sensitivity, a phenomenon where prices move against large orders before execution. The authors critique a prior approach by Church & Cliff (ISHV) that relies on top-of-book imbalance, demonstrating its fragility due to ignoring deeper limit order book (LOB) levels. They propose using Multi-Level Order Flow Imbalance (MLOFI), a metric that captures supply-demand imbalances across multiple LOB levels, providing a more robust signal. The authors extend three trading algorithms—ISHV, AA, and ZIP—with MLOFI, creating ZZISHV, ZZIAA, and ZZIZIP. Through simulations on the BSE platform, they show that these new agents reliably exhibit market impact effects, unlike the original algorithms. The work contributes improved impact-sensitive agents for both practical trading and realistic market simulation, with open-source code available.