MRGPRX1 plays dual roles in mediating nociception and pruritus, making it a promising target for alleviating itch and inhibiting pain. However, the mechanisms underlying MRGPRX1 activation and allosteric modulation remain poorly understood, posing significant challenges for structure-based drug discovery. Here, we employ Gaussian accelerated molecular dynamics (GaMD) for enhanced sampling of MRGPRX1, successfully obtaining a stable inactive conformation of the unbound receptor for the first time. Through comparative analyses with the previously resolved active conformation, we reveal how the subtle structural and dynamic changes of MRGPRX1 transition it from the inactive to the active state in the absence of the W6.48 toggle switch and other conserved motifs, providing insights into agonist-induced receptor activation. Furthermore, we integrate neural relational inference (NRI) deep learning with binding free energy calculations to elucidate the molecular basis of ML382's positive allosteric modulation (PAM) activity. Most strikingly, ML382 enhances the overall binding affinity of the peptide agonist BAM8-22 for MRGPRX1 via short-range interactions, while also modulates the conformations of the distal G-protein-binding site towards the fully active state through long-range pathways. Taken together, our study advances the molecular understanding of MRGPRX1 activation and allosteric modulation, thereby accelerating the rational design of analgesic and antipruritic drugs.