Summary: We present a new family of high-order shock-capturing finite difference numerical methods for systems of conservation laws. These methods, called Adaptive Compact Approximation Taylor (ACAT) schemes, use centered \((2 p + 1)\)-point stencils, where \(p\) may take values in \(\{1, 2, \dots, P \}\) according to a new family of smoothness indicators in the stencils. The methods are based on a combination of a robust first order scheme and the Compact Approximate Taylor (CAT) methods of order \(2p\)-order, \(p = 1, 2, \dots, P\) so that they are first order accurate near discontinuities and have order \(2p\) in smooth regions, where \((2 p + 1)\) is the size of the biggest stencil in which large gradients are not detected. CAT methods, introduced in [H. Carrillo and C. Parés, J. Sci. Comput. 80, No. 3, 1832–1866 (2019; Zbl 07121242)], are an extension to nonlinear problems of the Lax-Wendroff methods in which the Cauchy-Kovalesky (CK) procedure is circumvented following the strategy introduced in [D. Zorío et al., J. Sci. Comput. 71, No. 1, 246–273 (2017; Zbl 1387.65094)] that allows one to compute time derivatives in a recursive way using high-order centered differentiation formulas combined with Taylor expansions in time. The expression of ACAT methods for 1D and 2D systems of conservative laws is given and the performance is checked in a number of test cases for several linear and nonlinear systems of conservation laws, including Euler equations for gas dynamics.