We present a new algorithm for the identification and physical characterization of current sheets and reconnection sites in 2D and 3D large-scale relativistic magnetohydrodynamic numerical simulations. This has been implemented in the pluto code and tested in the cases of a single current sheet, a 2D jet, and a 3D unstable plasma column. Its main features are (i) a computational cost that allows its use in large-scale simulations and (ii) the capability to deal with complex 2D and 3D structures of the reconnection sites. In the performed simulations, we identify the computational cells that are part of a current sheet by a measure of the gradient of the magnetic field along different directions. Lagrangian particles, which follow the fluid, are used to sample plasma parameters before entering the reconnection sites that form during the evolution of the different configurations considered. Specifically, we track the distributions of the magnetization parameter sigma and the thermal to magnetic pressure ratio beta that - according to particle-in-cell simulation results - control the properties of particle acceleration in magnetic reconnection regions. Despite the fact that initial conditions of the simulations were not chosen 'ad hoc', the 3D simulation returns results suitable for efficient particle acceleration and realistic non-thermal particle distributions.
