SUMMARY Plant nucleotide-binding and leucine-rich repeat domain proteins (NLRs) are immune sensors that specifically recognize pathogen effectors and induce immune responses. Designing artificial NLRs with new effector recognition specificities is a promising prospect for sustainable, knowledge-driven crop protection. However, such strategies are hampered by the complexity of NLR function. Here, we tested whether molecular engineering of the integrated decoy domain (ID) of an NLR could extend its recognition spectrum to a new effector. To this aim, we relied on the detailed molecular knowledge of the recognition of distinct Magnaporthe oryzae MAX ( Magnaporthe AVRs and ToxB-like) effectors by the rice NLRs RGA5 and Pikp-1. For both NLRs, effector recognition involves physical binding to their HMA (Heavy Metal-Associated) IDs. However, AVR-PikD, the effector recognized by Pikp-1, binds to a completely different surface of the HMA domain compared to AVR-Pia and AVR1-CO39, recognized by RGA5. By introducing into the HMA domain of RGA5 the residues of the Pikp-1 HMA domain involved in AVR-PikD binding, we created a high-affinity binding surface for this new effector. In the Nicotiana benthamiana heterologous system, RGA5 variants carrying this engineered binding surface still recognize AVR-Pia and AVR1-CO39, but also perceive the new ligand, AVR-PikD, resulting in the activation of immune responses. Therefore, our study provides a proof of concept for the design of new effector recognition specificities in NLRs through molecular engineering of IDs. However, it pinpoints significant knowledge gaps that limit the full deployment of this NLR-ID engineering strategy and provides hypotheses for future research on this topic.