The goal of Hardening and Orchestrating Responses Under Stress (HORUS) project is to enable dynamic response and adaptive protection for cyber hunting scenarios, leveraging (1) a cognitive threat analysis process to choose from effective protective responses and actions, (2) a game-theoretic model for action selection and attack anticipation, and (3) adversarial analysis of threats and hardening of detectors. As a specific aspect of protective measures, the contributions of the project will explore novel application-level sensors and detectors that can modify the environment and deploy deceptive elements to contain, misdirect, and quarantine threats, while collecting critical threat intelligence and engage the attacker to interact with fictitious overlay networks, deeply integrated with the real production environment. This approach fundamentally differs from traditional countermeasures that aim at blocking any activities conforming to a specific rule, as it allows for deeper insights into threat actor characteristics and intents. Our research plan will be informed by several innovations that have been built as part of ongoing programs of all three proposer organization, namely the Cognitive Offense Analytics and automated workflows (IBM Research), Game-theoretic modeling for moving target defense (ASU), a Capture the Flag (CTF) as a service platform (ASU), protective measures and response platform (IBM Research), and adversarial machine learning analysis and protection (PSU/IBM Research).