<p>In this paper, we design an adaptive-gain super-twisting algorithm for a class of second-order nonlinear systems required to adhere to strict state constraints. The control gains are synthesized using barrier functions to ensure compliance with these constraints. Assuming known bound on the rate of the unknown disturbances, we introduce a norm observer to estimate a bounding function. This function is used to design the adaptive gains. We apply the proposed methodology to the problem of robust cooperative guidance in a three-body aircraft defense system, where the system is subject to line-of-sight angle constraints to enable timely and accurate combat response. The full nonlinear kinematics of the system, comprising a target, a defender, and an attacker, are considered. A key contribution lies in the construction of a novel norm observer, which incorporates minimal and practical knowledge of the unknown disturbance. Theoretical guarantees for the proposed control scheme are established using Lyapunov stability theory. Compared to existing methods, our design offers robust and continuous control with adaptively varying gains that enforce state constraints, while employing a computationally efficient norm observer. This leads to a significant reduction in control efforts. The effectiveness of the proposed approach is demonstrated through numerical simulations, involving a Van der Pol oscillator; as a motivating example of the nonlinear system, and the three-body aerial combat scenario.</p>